www.gusucode.com > nncontrol 工具箱 matlab 源码程序 > nncontrol/private/nnident.m
function nnident(cmd,arg1,arg2,arg3)
%NNIDENT Neural Network Identification GUI for Neural Network Controller Toolbox.
%
% Synopsis
%
% nnident(cmd,arg1,arg2,arg3)
%
% Warning!!
%
% This function may be altered or removed in future
% releases of Neural Network Toolbox. We recommend
% you do not write code which calls this function.
% This function is generally being called from a Simulink block.
% Orlando De Jesus, Martin Hagan, 1-25-00
% Copyright 1992-2011 The MathWorks, Inc.
% Orlando De Jesus. Strict generation of random signals for training.
% Final plots after training (Target and NN) have the same axis to easier comparative results.
% CONSTANTS
func_index=['trainbfg';'trainbr ';'traincgb';'traincgf';'traincgp';'traingd ';'traingdm';'traingda';'traingdx';'trainlm ';'trainoss';'trainrp ';'trainscg'];
% DEFAULTS
if nargin == 0, cmd = ''; else cmd = lower(cmd); end
% FIND WINDOW IF IT EXISTS
% 9/3/99 We alow the program to see hidden handles
fig = findall(0,'type','figure','tag','nnident');
if ~isempty(fig) && isempty(get(fig,'children')), fig = []; end
% GET WINDOW DATA IF IT EXISTS
if ~isempty(fig)
H = get(fig,'userdata');
if strcmp(cmd,'')
if get(H.gcbh_ptr,'userdata')~=arg1
delete(fig);
fig = [];
end
else
% ODJ 1-13-00 We check if the field SimulationStatus exist before reading that field
if isfield(get(H.gcbh_ptr),'UserData')
if isfield(get_param(get_param(get(H.gcbh_ptr,'userdata'),'parent'),'objectparameters'),'SimulationStatus')
SimulationStatus=get_param(get_param(get(H.gcbh_ptr,'userdata'),'parent'),'simulationstatus');
else
SimulationStatus='none';
end
else
SimulationStatus='none';
end
if (strcmp(SimulationStatus,'running') | strcmp(SimulationStatus,'paused')) & ~strcmp(cmd,'close')
set(H.error_messages,'string','You must stop the simulation to change NN configuration parameters.');
return;
end
end
end
%==================================================================
% Activate the window.
%
% ME() or ME('')
%==================================================================
if strcmp(cmd,'') | isempty(cmd)
if ~isempty(fig)
figure(fig)
set(fig,'visible','on')
else
nncontrolutil('nnident','init',arg1,arg2,arg3)
end
%==================================================================
% Close the window.
%
% ME() or ME('')
%==================================================================
elseif strcmp(cmd,'close') && ~isempty(fig)
arg1=get(H.gcbh_ptr,'userdata');
arg2=get(H.gcb_ptr,'userdata');
if exist(cat(2,tempdir,'nnidentdata.mat'))
delete(cat(2,tempdir,'nnidentdata.mat'));
end
parent_function=get(H.parent_function_ptr,'userdata');
if ~strcmp(parent_function,'narma_l2')
feval(parent_function,'',arg1,arg2,'nnident');
end
delete(fig);
% nnmodref('',arg1,arg2);
%==================================================================
% Initialize the window.
%
% ME('init')
%==================================================================
elseif strcmp(cmd,'init') && isempty(fig)
% 1-13-00 ODJ We check if the system is locked.
sys_par=arg2;
sys_par2=arg2;
while ~isempty(sys_par2)
sys_par=sys_par2;
sys_par2=get_param(sys_par,'parent');
end
if strcmp('on',get_param(sys_par,'lock'))
window_en='off';
else
window_en='on';
end
uipos = getuipos;
H.StdColor = get(0,'DefaultUicontrolBackgroundColor');
H.StdUnit='character';
H.PointsToPixels = 72/get(0,'ScreenPixelsPerInch');
if strcmp(arg3,'narma_l2')
H.me='Plant Identification - NARMA-L2';
else
H.me = 'Plant Identification';
end
fig = figure('Units',H.StdUnit, ...
'Color',[0.8 0.8 0.8], ...
'IntegerHandle', 'off',...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'MenuBar','none', ...
'Name',H.me, ...
'Numbertitle','off', ...
'Units', H.StdUnit, ...
'PaperUnits','points', ...
'Position',uipos.fig, ...
'Tag','nnident', ...
'Resize','off', ...
'ToolBar','none');
frame4 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'ListboxTop',0, ...
'Position',uipos.frame4, ...
'Style','frame', ...
'Tag','Frame4');
frame5 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'ListboxTop',0, ...
'Position',uipos.frame5, ...
'Style','frame', ...
'Tag','Frame5');
h1 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.h1_1, ...
'String','Training Parameters', ...
'Style','text', ...
'Tag','StaticText1');
frame1 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'ListboxTop',0, ...
'Position',uipos.frame1, ...
'Style','frame', ...
'Tag','Frame1');
h1 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.h1_2, ...
'String','Training Data', ...
'Style','text', ...
'Tag','StaticText1');
frame6 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'ListboxTop',0, ...
'Position',uipos.frame6, ...
'Style','frame', ...
'Tag','Frame6');
h1 = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.h1_3, ...
'String','Network Architecture', ...
'Style','text', ...
'Tag','StaticText1');
H.Title_nnident = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'FontSize',14, ...
'ListboxTop',0, ...
'Position',uipos.Title_nnident, ...
'String',H.me, ...
'Style','text', ...
'Tag','Title_nnident');
H.Hidden_layer_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Hidden_layer_text, ...
'String','Size of Hidden Layer', ...
'Style','text', ...
'ToolTipStr','Defines the size of the second layer of the neural network plant model.',...
'Tag','StaticText1');
H.Hidden_layer_size = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Hidden_layer_size, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Hidden_layer_size'', ''', get(H.Hidden_layer_text, 'String'),''');'], ...
'ToolTipStr','Defines the size of the second layer of the neural network plant model.',...
'Tag','Hidden_layer');
H.simulink_file_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.simulink_file_text, ...
'String','Simulink Plant Model:', ...
'HorizontalAlignment', 'right',...
'Style','text', ...
'ToolTipStr','Simulink file containing the plant to be modeled.',...
'Tag','StaticText1');
H.BrowseButton = uicontrol('Parent',fig, ...
'Unit',H.StdUnit, ...
'Callback','nncontrolutil(''nnident'',''browsesim'',gcbf);', ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position', uipos.BrowseButton, ...
'String','Browse', ...
'ToolTipStr','Allow the user to select a Simulink file.',...
'Tag','BrowseButton');
H.simulink_file = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'Callback','nncontrolutil(''nnident'',''clearpath'',gcbf);', ...
'HorizontalAlignment','left', ...
'ListboxTop',0, ...
'Position',uipos.simulink_file, ...
'Style','edit', ...
'ToolTipStr','Simulink file containing the plant to be modeled.',...
'Tag','Plant_model');
H.Sampling_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Sampling_text, ...
'String','Sampling Interval (sec) ', ...
'Style','text', ...
'ToolTipStr','Sampling interval at which the data will be collected from the Simulink plant model.',...
'Tag','StaticText1');
H.Sampling_time = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Sampling_time, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Sampling_time'', ''', get(H.Sampling_text, 'String'),''');'], ...
'ToolTipStr','Sampling interval at which the data will be collected from the Simulink plant model.',...
'Tag','Sampling_time');
H.Delayed_input_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Delayed_input_text, ...
'String','No. Delayed Plant Inputs', ...
'Style','text', ...
'ToolTipStr','Defines how many delays on the plant input will be used to feed the neural network plant model.',...
'Tag','StaticText1');
H.Delayed_input = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Max',500, ...
'Min',1, ...
'Position',uipos.Delayed_input, ...
'Style','edit', ...
'Tag','Ni', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Delayed_input'', ''', get(H.Delayed_input_text, 'String'),''');'], ...
'ToolTipStr','Defines how many delays on the plant input will be used to feed the neural network plant model.',...
'Value',1);
H.Delayed_output_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Delayed_output_text, ...
'String','No. Delayed Plant Outputs', ...
'Style','text', ...
'ToolTipStr','Defines how many delays on the plant output will be used to feed the neural network plant model.',...
'Tag','StaticText1');
H.Delayed_output = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Max',500, ...
'Min',1, ...
'Position',uipos.Delayed_output, ...
'Style','edit', ...
'Tag','Nj', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Delayed_output'', ''', get(H.Delayed_output_text, 'String'),''');'], ...
'ToolTipStr','Defines how many delays on the plant output will be used to feed the neural network plant model.',...
'Value',1);
H.Limit_output_data = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Callback','nncontrolutil(''nnident'',''limit_output'')', ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Limit_output_data, ...
'String','Limit Output Data', ...
'Style','checkbox', ...
'Tag','checkbox3', ...
'ToolTipStr','If selected, the plant output data will be bounded.',...
'Value',1);
H.Normalize_data = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Normalize_data, ...
'String','Normalize Training Data', ...
'Style','checkbox', ...
'Tag','checkbox2', ...
'ToolTipStr','If selected, the plant input-output data will be normalized.',...
'Value',1);
H.Use_Previous_Weights_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Use_Previous_Weights_but, ...
'String','Use Current Weights', ...
'Style','checkbox', ...
'Tag','checkbox2', ...
'ToolTipStr','If selected, the current weights are used as initial values for continued training.',...
'Value',0);
H.Use_Validation_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Use_Validation_but, ...
'String','Use Validation Data', ...
'Style','checkbox', ...
'ToolTipStr','A validation data set will be used to stop training.',...
'Tag','checkbox1');
H.Use_Testing_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Use_Testing_but, ...
'String','Use Testing Data', ...
'Style','checkbox', ...
'ToolTipStr','A testing data set will be monitored during training.',...
'Tag','checkbox1a');
H.Max_input_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Max_input_text, ...
'String','Maximum Plant Input', ...
'Style','text', ...
'ToolTipStr','Defines an upper bound on the random plant input.',...
'Tag','Maximum_input_Text');
H.Max_input = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Max_input, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Max_input'', ''', get(H.Max_input_text, 'String'),''');'], ...
'ToolTipStr','Defines an upper bound on the random plant input.',...
'Tag','Maximum_input');
H.Min_input_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Min_input_text, ...
'String','Minimum Plant Input', ...
'Style','text', ...
'ToolTipStr','Defines a lower bound on the random plant input.',...
'Tag','Minimum_input_Text');
H.Min_input = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Min_input, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Min_input'', ''', get(H.Min_input_text, 'String'),''');'], ...
'ToolTipStr','Defines a lower bound on the random plant input.',...
'Tag','Minimum_input');
H.max_int_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.max_int_text, ...
'String','Maximum Interval Value (sec) ', ...
'Style','text', ...
'ToolTipStr','Defines a maximum interval over which the random plant input will remain constant.',...
'Tag','StaticText2');
H.max_int_edit = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.max_int_edit, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''max_int_edit'', ''', get(H.max_int_text, 'String'),''');'], ...
'ToolTipStr','Defines a maximum interval over which the random plant input will remain constant.',...
'Tag','max_r_edit');
H.min_int_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.min_int_text, ...
'String','Minimum Interval Value (sec) ', ...
'Style','text', ...
'ToolTipStr','Defines a minimum interval over which the random plant input will remain constant.',...
'Tag','StaticText2');
H.min_int_edit = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.min_int_edit, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''min_int_edit'', ''', get(H.min_int_text, 'String'),''');'], ...
'ToolTipStr','Defines a minimum interval over which the random plant input will remain constant.',...
'Tag','EditText1');
H.Max_output_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Max_output_text, ...
'String','Maximum Plant Output', ...
'Style','text', ...
'ToolTipStr','Defines an upper bound on the plant output.',...
'Tag','Maximum_output_text');
H.Max_output = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Max_output, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Max_output'', ''', get(H.Max_output_text, 'String'),''' ,false);'], ...
'ToolTipStr','Defines an upper bound on the plant output.',...
'Tag','Maximum_output');
H.Min_output_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Min_output_text, ...
'String','Minimum Plant Output', ...
'Style','text', ...
'ToolTipStr','Defines a lower bound on the plant output.',...
'Tag','Minimum_output_text');
H.Min_output = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Min_output, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Min_output'', ''', get(H.Min_output_text, 'String'),''' ,false);'], ...
'ToolTipStr','Defines a lower bound on the plant output.',...
'Tag','Minimum_output');
H.Samples_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.Samples_text, ...
'String','Training Samples', ...
'Style','text', ...
'ToolTipStr','Defines how many data points will be generated for training.',...
'Tag','StaticText1');
H.Samples = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Samples, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''Samples'', ''', get(H.Samples_text, 'String'),''');'], ...
'ToolTipStr','Defines how many data points will be generated for training.',...
'Tag','Training_examples');
H.trainfun_edit = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.trainfun_edit, ...
'Max',17, ...
'String',['trainbfg';'trainbr ';'traincgb';'traincgf';'traincgp';'traingd ';'traingdm';'traingda';'traingdx';'trainlm ';'trainoss';'trainrp ';'trainscg'], ...
'Style','popupmenu', ...
'Tag','PopupMenu1', ...
'Value',1);
H.trainfun_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.trainfun_text, ...
'String','Training Function', ...
'Style','text', ...
'ToolTipStr','Select a training function for neural network plant model training.',...
'Tag','StaticText3');
H.epochs_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'Enable',window_en, ...
'HorizontalAlignment','right', ...
'ListboxTop',0, ...
'Position',uipos.epochs_text, ...
'String','Training Epochs', ...
'Style','text', ...
'ToolTipStr','Defines number of iterations to train the neural network plant model.',...
'Tag','StaticText1');
H.epochs_h = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[1 1 1], ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.epochs_h, ...
'Style','edit', ...
'Callback',['nncontrolutil(''nnident'',''check_params'',''epochs_h'', ''', get(H.epochs_text, 'String'),''');'], ...
'ToolTipStr','Defines number of iterations to train the neural network plant model.',...
'Tag','Training_epochs');
H.Start_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''continue_training'')', ...
'Enable','off', ...
'ListboxTop',0, ...
'Position',uipos.Start_but, ...
'String','Train Network', ...
'ToolTipStr','Train the neural network plant using the parameters shown in this window.',...
'Tag','Pushbutton1');
H.OK_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''ok'')', ...
'Enable','off', ...
'ListboxTop',0, ...
'Position',uipos.OK_but, ...
'String','OK', ...
'ToolTipStr','Save the neural network plant model into the controller block and close this menu.',...
'Tag','Pushbutton1');
H.Cancel_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''close'')', ...
'ListboxTop',0, ...
'Position',uipos.Cancel_but, ...
'String','Cancel', ...
'ToolTipStr','Discard the neural network plant model and close this menu',...
'Tag','Pushbutton1');
H.Apply_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''apply'')', ...
'Enable','off', ...
'ListboxTop',0, ...
'Position',uipos.Apply_but, ...
'String','Apply', ...
'ToolTipStr','Save the neural network plant model into the controller block',...
'Tag','Pushbutton1');
H.Simulating_text = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.8 0.8 0.8], ...
'FontWeight','bold', ...
'ListboxTop',0, ...
'Position',uipos.Simulating_text, ...
'String','Simulating Plant', ...
'Style','text', ...
'Tag','StaticText1', ...
'visible','off');
H.error_messages= uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'FontWeight','bold', ...
'ForegroundColor',[0 0 1], ...
'ListboxTop',0, ...
'Position',uipos.error_messages, ...
'Style','text', ...
'ToolTipStr','Feedback line with important messages for the user.',...
'Tag','StaticText1');
H.Gen_data_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''gen_data'')', ...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Gen_data_but, ...
'String','Generate Training Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Generate data to be used in training the neural network plant model.');
if strcmp(arg3,'nnpredict')
H.Get_data_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataimport'',''init'',gcbf,''nnpredict'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Get_data_file_but, ...
'String','Import Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Import training data from the workspace or from a file.');
H.Save_to_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataexport'',''init'',gcbf,''nnpredict'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Save_to_file_but, ...
'String','Export Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Export training data to the workspace or to a file.');
elseif strcmp(arg3,'nnmodref')
H.Get_data_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataimport'',''init'',gcbf,''nnmodref'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Get_data_file_but, ...
'String','Import Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Import training data from the workspace or from a file.');
H.Save_to_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataexport'',''init'',gcbf,''nnmodref'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Save_to_file_but, ...
'String','Export Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Export training data to the workspace or to a file.');
elseif strcmp(arg3,'narma_l2')
H.Get_data_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataimport'',''init'',gcbf,''narma_l2'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Get_data_file_but, ...
'String','Import Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Import training data from the workspace or from a file.');
H.Save_to_file_but = uicontrol('Parent',fig, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nndataexport'',''init'',gcbf,''narma_l2'');',...
'Enable',window_en, ...
'ListboxTop',0, ...
'Position',uipos.Save_to_file_but, ...
'String','Export Data', ...
'Tag','Pushbutton1', ...
'TooltipString','Export training data to the workspace or to a file.');
end
H.Training_done=0;
H.Data_Generated=1;
H.Data_Available=0;
H.Data_Imported=0;
% We create the menus for the block.
H.Handles.Menus.File.Top= uimenu('Parent',fig, ...
'Label','File');
if strcmp(arg3,'nnpredict')
H.Handles.Menus.File.ImportModel = uimenu('Parent',...
H.Handles.Menus.File.Top,...
'Label','Import Network...',...
'Accelerator','I',...
'Callback','nncontrolutil(''nnimport'',''init'',gcbf,''nnpredict'',''nnident'');',...
'Enable',window_en, ...
'Tag','ImportModel');
H.Handles.Menus.File.Export = uimenu('Parent',H.Handles.Menus.File.Top, ...
'Label','Export Network...', ...
'Accelerator','E', ...
'Callback','nncontrolutil(''nnexport'',''init'',gcbf,''nnpredict'',''nnident'')', ...
'Enable',window_en, ...
'Tag','ExportMenu');
elseif strcmp(arg3,'nnmodref')
H.Handles.Menus.File.ImportModel = uimenu('Parent',...
H.Handles.Menus.File.Top,...
'Label','Import Network...',...
'Accelerator','I',...
'Callback','nncontrolutil(''nnimport'',''init'',gcbf,''nnmodref'',''nnident'');',...
'Enable',window_en, ...
'Tag','ImportModel');
H.Handles.Menus.File.Export = uimenu('Parent',H.Handles.Menus.File.Top, ...
'Label','Export Network...', ...
'Accelerator','E', ...
'Callback','nncontrolutil(''nnexport'',''init'',gcbf,''nnmodref'',''nnident'')', ...
'Enable',window_en, ...
'Tag','ExportMenu');
elseif strcmp(arg3,'narma_l2')
H.Handles.Menus.File.ImportModel = uimenu('Parent',...
H.Handles.Menus.File.Top,...
'Label','Import Network...',...
'Accelerator','I',...
'Callback','nncontrolutil(''nnimport'',''init'',gcbf,''narma_l2'',''nnident'');',...
'Enable',window_en, ...
'Tag','ImportModel');
H.Handles.Menus.File.Export = uimenu('Parent',H.Handles.Menus.File.Top, ...
'Label','Export Network...', ...
'Accelerator','E', ...
'Callback','nncontrolutil(''nnexport'',''init'',gcbf,''narma_l2'',''nnident'')', ...
'Enable',window_en, ...
'Tag','ExportMenu');
end
H.Handles.Menus.File.Save_NN = uimenu('Parent',...
H.Handles.Menus.File.Top,...
'Label','Save',...
'Separator','on', ...
'Enable','off', ...
'Accelerator','S',...
'Callback','nncontrolutil(''nnident'',''apply'');',...
'Tag','ImportModel');
H.Handles.Menus.File.Save_Exit_NN = uimenu('Parent',...
H.Handles.Menus.File.Top,...
'Label','Save and Exit',...
'Enable','off', ...
'Accelerator','A',...
'Callback','nncontrolutil(''nnident'',''ok'');',...
'Tag','ImportModel');
H.Handles.Menus.File.Close = uimenu('Parent',H.Handles.Menus.File.Top, ...
'Callback','nncontrolutil(''nnident'',''close'',gcbf);', ...
'Separator','on', ...
'Label','Exit without saving', ...
'Accelerator','X', ...
'Tag','CloseMenu');
H.Handles.Menus.Window.Top = matlab.ui.internal.createWinMenu(fig);
H.Handles.Menus.Help.Top = uimenu('Parent',fig, ...
'Label','Help');
H.Handles.Menus.Help.Main = uimenu('Parent',H.Handles.Menus.Help.Top, ...
'Label','Main Help', ...
'Callback','nncontrolutil(''nnidenthelp'',''main'');',...
'Accelerator','H');
H.Handles.Menus.Help.PlantIdent = uimenu('Parent',H.Handles.Menus.Help.Top, ...
'Label','Plant Identification...', ...
'Separator','on',...
'CallBack','nncontrolutil(''nnidenthelp'',''plant_ident'');');
H.gcbh_ptr = uicontrol('Parent',fig,'visible','off');
set(H.gcbh_ptr,'userdata',arg1);
H.gcb_ptr = uicontrol('Parent',fig,'visible','off');
set(H.gcb_ptr,'userdata',arg2);
S1=convertEmptyBrackets(get_param(arg1,'S1')); % S1 is ASCII
if isempty(S1) % If the field is empty we initialize default value.
S1=num2str(0);
else
set(H.Hidden_layer_size,'string',S1);
end
H.S1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.S1_ptr,'userdata',str2num(S1)); % S1 is saved as number
sim_file=convertEmptyBrackets(get_param(arg1,'sim_file'));
if isempty(sim_file) % If the field is empty we initialize default value.
sim_file='';
end
H.sim_file_ptr = uicontrol('Parent',fig,'visible','off');
set(H.simulink_file,'string',sim_file,'UserData',struct('FileName',sim_file,'PathName',[]));
set(H.sim_file_ptr,'userdata',sim_file);
Ts=convertEmptyBrackets(get_param(arg1,'Ts'));
if isempty(Ts) % If the field is empty we initialize default value.
Ts=num2str(0);
else
set(H.Sampling_time,'string',Ts);
end
H.Ts_ptr = uicontrol('Parent',fig,'visible','off');
set(H.Ts_ptr,'userdata',str2num(Ts));
Ni=convertEmptyBrackets(get_param(arg1,'Ni'));
if isempty(Ni) % If the field is empty we initialize default value.
Ni=num2str(0);
else
set(H.Delayed_input,'string',Ni);
end
H.Ni_ptr = uicontrol('Parent',fig,'visible','off');
set(H.Ni_ptr,'userdata',str2num(Ni));
Nj=convertEmptyBrackets(get_param(arg1,'Nj'));
if isempty(Nj) % If the field is empty we initialize default value.
Nj=num2str(0);
else
set(H.Delayed_output,'string',Nj);
end
H.Nj_ptr = uicontrol('Parent',fig,'visible','off');
set(H.Nj_ptr,'userdata',str2num(Nj));
trainfun=get_param(arg1,'trainfun');
if size(trainfun,2)==6
trainfun(7:8)=' ';
elseif size(trainfun,2)==7
trainfun(8)=' ';
end
vv=nnstring.first_match(trainfun,func_index);
set(H.trainfun_edit,'value',vv);
Use_Validation=convertEmptyBrackets(get_param(arg1,'Use_Validation'));
if isempty(Use_Validation) % If the field is empty we initialize default value.
Use_Validation=num2str(1);
end
set(H.Use_Validation_but,'value',str2num(Use_Validation));
Use_Testing=convertEmptyBrackets(get_param(arg1,'Use_Testing'));
if isempty(Use_Testing) % If the field is empty we initialize default value.
Use_Testing=num2str(1);
end
set(H.Use_Testing_but,'value',str2num(Use_Testing))
max_i=convertEmptyBrackets(get_param(arg1,'max_i'));
if isempty(max_i) % If the field is empty we initialize default value.
max_i=num2str(0);
else
set(H.Max_input,'string',max_i);
end
H.max_i_ptr = uicontrol('Parent',fig,'visible','off');
set(H.max_i_ptr,'userdata',str2num(max_i));
min_i=convertEmptyBrackets(get_param(arg1,'min_i'));
if isempty(min_i) % If the field is empty we initialize default value.
min_i=num2str(0);
else
set(H.Min_input,'string',min_i);
end
H.min_i_ptr = uicontrol('Parent',fig,'visible','off');
set(H.min_i_ptr,'userdata',str2num(min_i));
max_i_int=convertEmptyBrackets(get_param(arg1,'max_i_int'));
if isempty(max_i_int) % If the field is empty we initialize default value.
max_i_int=num2str(0);
else
set(H.max_int_edit,'string',max_i_int);
end
H.max_i_int_ptr = uicontrol('Parent',fig,'visible','off');
set(H.max_i_int_ptr,'userdata',str2num(max_i_int));
min_i_int=convertEmptyBrackets(get_param(arg1,'min_i_int'));
if isempty(min_i_int) % If the field is empty we initialize default value.
min_i_int=num2str(0);
else
set(H.min_int_edit,'string',min_i_int);
end
H.min_i_int_ptr = uicontrol('Parent',fig,'visible','off');
set(H.min_i_int_ptr,'userdata',str2num(min_i_int));
Limit_output=convertEmptyBrackets(get_param(arg1,'limit_output'));
if isempty(Limit_output)
Limit_output='0';
end
set(H.Limit_output_data,'value',str2num(Limit_output))
max_out=convertEmptyBrackets(get_param(arg1,'max_output'));
if isempty(max_out) % If the field is empty we initialize default value.
max_out=num2str(0);
else
set(H.Max_output,'string',max_out);
end
H.max_out_ptr = uicontrol('Parent',fig,'visible','off');
set(H.max_out_ptr,'userdata',str2num(max_out));
min_out=convertEmptyBrackets(get_param(arg1,'min_output'));
if isempty(min_out) % If the field is empty we initialize default value.
min_out=num2str(0);
else
set(H.Min_output,'string',min_out);
end
H.min_out_ptr = uicontrol('Parent',fig,'visible','off');
set(H.min_out_ptr,'userdata',str2num(min_out));
Use_Previous_Weights=convertEmptyBrackets(get_param(arg1,'Use_Previous_Weights'));
if isempty(Use_Previous_Weights) % If the field is empty we initialize default value.
Use_Previous_Weights=num2str(1);
end
set(H.Use_Previous_Weights_but,'value',str2num(Use_Previous_Weights));
H.Use_Previous_Weights_ptr = uicontrol('Parent',fig,'visible','off');
set(H.Use_Previous_Weights_ptr,'userdata',str2num(Use_Previous_Weights));
sam_training=convertEmptyBrackets(get_param(arg1,'sam_training'));
if isempty(sam_training) % If the field is empty we initialize default value.
sam_training=num2str(0);
else
set(H.Samples,'string',sam_training);
end
H.sam_training_ptr = uicontrol('Parent',fig,'visible','off');
set(H.sam_training_ptr,'userdata',str2num(sam_training));
epochs=convertEmptyBrackets(get_param(arg1,'epochs'));
if isempty(epochs) % If the field is empty we initialize default value.
epochs=num2str(0);
else
set(H.epochs_h,'string',epochs);
end
H.epochs_ptr = uicontrol('Parent',fig,'visible','off');
set(H.epochs_ptr,'userdata',str2num(epochs));
if strcmp(arg3,'narma_l2')
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
IW1_1=eval(nnstring.vert_cat(get_param(arg1,'IW1_1')),'0');
H.IW1_1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.IW1_1_ptr,'userdata',IW1_1);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
IW3_2=eval(nnstring.vert_cat(get_param(arg1,'IW3_2')),'0');
H.IW3_2_ptr = uicontrol('Parent',fig,'visible','off');
set(H.IW3_2_ptr,'userdata',IW3_2);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
IW5_3=eval(nnstring.vert_cat(get_param(arg1,'IW5_3')),'0');
H.IW5_3_ptr = uicontrol('Parent',fig,'visible','off');
set(H.IW5_3_ptr,'userdata',IW5_3);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW2_1=eval(nnstring.vert_cat(get_param(arg1,'LW2_1')),'0');
H.LW2_1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW2_1_ptr,'userdata',LW2_1);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW4_3=eval(nnstring.vert_cat(get_param(arg1,'LW4_3')),'0');
H.LW4_3_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW4_3_ptr,'userdata',LW4_3);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW5_4=eval(nnstring.vert_cat(get_param(arg1,'LW5_4')),'0');
H.LW5_4_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW5_4_ptr,'userdata',LW5_4);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW6_5=eval(nnstring.vert_cat(get_param(arg1,'LW6_5')),'0');
H.LW6_5_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW6_5_ptr,'userdata',LW6_5);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW6_2=eval(nnstring.vert_cat(get_param(arg1,'LW6_2')),'0');
H.LW6_2_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW6_2_ptr,'userdata',LW6_2);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B1=eval(nnstring.vert_cat(get_param(arg1,'B1')),'0');
H.B1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B1_ptr,'userdata',B1);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B2=eval(nnstring.vert_cat(get_param(arg1,'B2')),'0');
H.B2_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B2_ptr,'userdata',B2);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B3=eval(nnstring.vert_cat(get_param(arg1,'B3')),'0');
H.B3_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B3_ptr,'userdata',B3);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B4=eval(nnstring.vert_cat(get_param(arg1,'B4')),'0');
H.B4_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B4_ptr,'userdata',B4);
else
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
IW=eval(nnstring.vert_cat(get_param(arg1,'IW')),'0');
H.IW_ptr = uicontrol('Parent',fig,'visible','off');
set(H.IW_ptr,'userdata',IW);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW1_2=eval(nnstring.vert_cat(get_param(arg1,'LW1_2')),'0');
H.LW1_2_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW1_2_ptr,'userdata',LW1_2);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
LW2_1=eval(nnstring.vert_cat(get_param(arg1,'LW2_1')),'0');
H.LW2_1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.LW2_1_ptr,'userdata',LW2_1);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B1=eval(nnstring.vert_cat(get_param(arg1,'B1')),'0');
H.B1_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B1_ptr,'userdata',B1);
% ODJ 1-13-00 Default values are assigned if the weight fields are empty.
B2=eval(nnstring.vert_cat(get_param(arg1,'B2')),'0');
H.B2_ptr = uicontrol('Parent',fig,'visible','off');
set(H.B2_ptr,'userdata',B2);
end
minp=get_param(arg1,'minp');
H.minp_ptr = uicontrol('Parent',fig,'visible','off');
set(H.minp_ptr,'userdata',minp);
maxp=get_param(arg1,'maxp');
H.maxp_ptr = uicontrol('Parent',fig,'visible','off');
set(H.maxp_ptr,'userdata',maxp);
mint=get_param(arg1,'mint');
H.mint_ptr = uicontrol('Parent',fig,'visible','off');
set(H.mint_ptr,'userdata',mint);
maxt=get_param(arg1,'maxt');
H.maxt_ptr = uicontrol('Parent',fig,'visible','off');
set(H.maxt_ptr,'userdata',maxt);
Normalize=str2num(get_param(arg1,'Normalize'));
if isempty(Normalize) % If the field is empty we initialize default value.
Normalize=0;
end
H.Normalize_ptr = uicontrol('Parent',fig,'visible','off');
set(H.Normalize_ptr,'userdata',Normalize);
set(H.Normalize_data,'value',Normalize);
H.In_training_ptr = uicontrol('Parent',fig,'visible','off');
set(H.In_training_ptr,'userdata',0);
H.parent_function_ptr = uicontrol('Parent',fig,'visible','off');
set(H.parent_function_ptr,'userdata',arg3);
set(fig,'userdata',H);
set(H.error_messages,'string',sprintf('Generate or import data before training the neural network plant.'));
% IMPORTANT: This call must be here because we need H for the callback.
nncontrolutil('nnident','limit_output');
elseif strcmp(cmd,'limit_output') && ~isempty(fig)
limit_output=get(H.Limit_output_data,'Value');
if limit_output==1
set(H.Max_output,'enable','on')
set(H.Max_output_text,'enable','on')
set(H.Min_output,'enable','on')
set(H.Min_output_text,'enable','on')
else
set(H.Max_output,'enable','off')
set(H.Max_output_text,'enable','off')
set(H.Min_output,'enable','off')
set(H.Min_output_text,'enable','off')
end
elseif strcmp(cmd,'data_no_ok') && ~isempty(fig)
% set(H.Start_but,'enable','on')
set(H.Cancel_but,'enable','on')
if H.Training_done
set(H.OK_but,'enable','on')
set(H.Apply_but,'enable','on')
set(H.Handles.Menus.File.Save_NN,'enable','on')
set(H.Handles.Menus.File.Save_Exit_NN,'enable','on')
end
if H.Data_Available
load(cat(2,tempdir,'nnidentdata.mat'),'N2');
set(H.Start_but,'enable','on')
st=sprintf('Your training data set has %d samples.\nYou can now train the network.',N2-1);
set(H.error_messages,'string',st);
else
set(H.error_messages,'string',sprintf('Generate or import data before training the neural network plant.'));
set(H.Max_input,'enable','on')
set(H.Max_input_text,'enable','on')
set(H.Min_input,'enable','on')
set(H.Min_input_text,'enable','on')
set(H.max_int_edit,'enable','on')
set(H.max_int_text,'enable','on')
set(H.min_int_edit,'enable','on')
set(H.min_int_text,'enable','on')
set(H.Samples_text,'enable','on')
set(H.Samples,'enable','on')
set(H.Sampling_text,'enable','on')
set(H.Sampling_time,'enable','on')
set(H.Limit_output_data,'enable','on');
limit_output=get(H.Limit_output_data,'Value');
if limit_output==1
set(H.Max_output,'enable','on')
set(H.Max_output_text,'enable','on')
set(H.Min_output,'enable','on')
set(H.Min_output_text,'enable','on')
end
set(H.BrowseButton,'enable','on');
set(H.simulink_file,'enable','on');
set(H.simulink_file_text,'enable','on');
end
set(H.Simulating_text,'visible','off'); drawnow; % pause needed to refresh the message
fig2=findall(0,'type','figure','tag','nnidentdata');
if ~isempty(fig2)
delete(fig2);
end
if exist(cat(2,tempdir,'nnidentdata2.mat'))
delete(cat(2,tempdir,'nnidentdata2.mat'));
end
% We refresh the menu.
arg1=get(H.gcbh_ptr,'userdata');
arg2=get(H.gcb_ptr,'userdata');
nncontrolutil('nnident','',arg1,arg2,'');
elseif strcmp(cmd,'browsesim')
filterspec = '*.mdl;*.slx';
udFileEdit = get(H.simulink_file,'UserData');
LastPath = udFileEdit.PathName;
CurrentPath=pwd;
if ~isempty(LastPath),
cd(LastPath);
end
[filename,pathname] = uigetfile(filterspec,'Simulink Plant Model:');
if ~isempty(LastPath),
cd(CurrentPath);
end
if filename,
if ~strcmpi(pathname(1:end-1),CurrentPath)
ImportStr = [pathname,filename(1:end-4)];
else
ImportStr = filename(1:end-4);
end
udFileEdit.PathName=pathname;
udFileEdit.FileName=filename;
set(H.simulink_file,'String',filename(1:end-4),'UserData',udFileEdit);
end
elseif strcmp(cmd,'clearpath') && ~isempty(fig)
%---Callback for the SImulink File box
% Whenever a new name is entered, update the Userdata
NewName = get(gcbo,'String');
indDot = strfind(NewName,'.');
if ~isempty(indDot),
NewName=NewName(1:indDot(end)-1);
set(H.simulink_file,'String',NewName)
end
elseif strcmp(cmd,'erase_data') && ~isempty(fig)
set(H.Max_input,'enable','on')
set(H.Max_input_text,'enable','on')
set(H.Min_input,'enable','on')
set(H.Min_input_text,'enable','on')
set(H.max_int_edit,'enable','on')
set(H.max_int_text,'enable','on')
set(H.min_int_edit,'enable','on')
set(H.min_int_text,'enable','on')
set(H.Samples_text,'enable','on')
set(H.Samples,'enable','on')
set(H.Sampling_text,'enable','on')
set(H.Sampling_time,'enable','on')
set(H.Limit_output_data,'enable','on');
limit_output=get(H.Limit_output_data,'Value');
if limit_output==1
set(H.Max_output,'enable','on')
set(H.Max_output_text,'enable','on')
set(H.Min_output,'enable','on')
set(H.Min_output_text,'enable','on')
end
set(H.BrowseButton,'enable','on');
set(H.simulink_file,'enable','on');
set(H.simulink_file_text,'enable','on');
H.Data_Generated=0;
H.Data_Imported=0;
H.Data_Available=0;
set(H.Start_but,'enable','off')
if exist(cat(2,tempdir,'nnidentdata2.mat'),'file')
delete(cat(2,tempdir,'nnidentdata2.mat'));
end
if exist(cat(2,tempdir,'nnidentdata.mat'),'file')
delete(cat(2,tempdir,'nnidentdata.mat'));
end
set(fig,'UserData',H);
set(H.Gen_data_but,'String','Generate Training Data', ...
'Callback','nncontrolutil(''nnident'',''gen_data'')', ...
'TooltipString','Generate data to be used in training the neural network plant model.');
set(H.error_messages,'string',sprintf('Generate or import data before training the neural network plant.'));
elseif (strcmp(cmd,'start_training') | strcmp(cmd,'continue_training') | strcmp(cmd,'data_ok') | ...
strcmp(cmd,'gen_data') | strcmp(cmd,'have_file')) && ~isempty(fig)
if strcmp(cmd,'gen_data') && ~isempty(fig)
% H.Data_Generated=1;
H.Data_Imported=0;
% H.Training_done=0;
set(fig,'UserData',H);
elseif strcmp(cmd,'have_file') && ~isempty(fig)
ImportStr=arg1;
H.Data_Imported=1;
if nargin==3
Data_Name=arg2;
else
U_Name=arg2;
Y_Name=arg3;
end
end
set(H.Start_but,'enable','off')
set(H.Cancel_but,'enable','off')
set(H.OK_but,'enable','off')
set(H.Apply_but,'enable','off')
set(H.Handles.Menus.File.Save_NN,'enable','off')
set(H.Handles.Menus.File.Save_Exit_NN,'enable','off')
if (strcmp(cmd,'gen_data') | strcmp(cmd,'have_file'))%strcmp(cmd,'start_training')
arg1=get(H.gcbh_ptr,'userdata');
a1 = str2num(get(H.Sampling_time,'string'));
Ts=get_param(arg1,'Ts');
if length(a1) == 0,
present_error(fig,H,H.Sampling_time,Ts,1, ...
'You must initialize the sampling interval of your plant before training the neural network');
return
elseif a1<=0 | ~sanitycheckparam(a1),
present_error(fig,H,H.Sampling_time,Ts,1, ...
'You must set a positive sampling interval of your plant before training the neural network');
return
else Ts=a1; set(H.Ts_ptr,'userdata',Ts); end
a1 = str2num(get(H.Max_input,'string'));
max_i=get_param(arg1,'max_i');
if ~sanitycheckparam(a1),
present_error(fig,H,H.Max_input,max_i,1, ...
'You must enter a valid number for the maximum plant input.');
return
else max_i=a1; set(H.max_i_ptr,'userdata',max_i); end
a1 = str2num(get(H.Min_input,'string'));
min_i=get_param(arg1,'min_i');
if ~sanitycheckparam(a1),
present_error(fig,H,H.Min_input,min_i,1, ...
'You must enter a valid number for the minimum plant input.');
return
elseif a1>=max_i
present_error(fig,H,H.Min_input,min_i,1, ...
'You must enter valid numbers for the maximum and minimum plant inputs.');
return
else min_i=a1; set(H.min_i_ptr,'userdata',min_i); end
fig2 = findall(0,'type','figure','tag','nnidentdata');
if strcmp(cmd,'have_file')
if nargin==3 % Structure
if isempty(ImportStr) % Workspace
tr_dat=evalin('base',Data_Name);
if ~isfield(tr_dat,'flag')
tr_dat.flag=ones(size(tr_dat.Y));
end
if ~isfield(tr_dat,'Ts')
tr_dat.Ts=Ts;
end
else
a1 = ImportStr;
a2 = which(cat(2,a1,'.mat'));
if (isempty(a1) || isempty(a2)),
present_error(fig,H,0,0,0, ...
'You must enter a valid filename for your training data, or the file directory must be defined in the MATLAB Path.');
return
else
file_data=a1;
end
temp=load (file_data,Data_Name);
tr_dat.U=getfield(temp,Data_Name,'U');
tr_dat.Y=getfield(temp,Data_Name,'Y');
if isfield(eval(cat(2,'temp.',Data_Name)),'flag')
tr_dat.flag=getfield(temp,Data_Name,'flag');
else
tr_dat.flag=ones(size(tr_dat.Y));
end
if isfield(eval(cat(2,'temp.',Data_Name)),'Ts')
tr_dat.Ts=getfield(temp,Data_Name,'Ts');
else
tr_dat.Ts=Ts;
end
end
else % Arrays.
if isempty(ImportStr) % Workspace
tr_dat=struct('U',evalin('base',U_Name),'Y',evalin('base',Y_Name));
tr_dat.flag=ones(size(tr_dat.Y));
tr_dat.Ts=Ts;
else
a1 = ImportStr;
a2 = which(cat(2,a1,'.mat'));
if (length(a1) == 0 | length(a2) == 0),
present_error(fig,H,0,0,0, ...
'You must enter a valid filename for your training data, or the file directory must be defined in the MATLAB Path.');
return
else file_data=a1; end
temp=load (file_data,U_Name,Y_Name);
tr_dat.U=getfield(temp,U_Name);
tr_dat.Y=getfield(temp,Y_Name);
tr_dat.flag=ones(size(tr_dat.Y));
tr_dat.Ts=Ts;
end
end
% We verify direction of the input vectors.
if size(tr_dat.U,1)<=1
tr_dat.U=tr_dat.U';
end
if size(tr_dat.Y,1)<=1
tr_dat.Y=tr_dat.Y';
end
if size(tr_dat.flag,1)<=1
tr_dat.flag=tr_dat.flag';
end
sam_training=size(tr_dat.Y,1)-1;
uipos = getuipos;
if isempty(fig2)
pos_fig2=get(fig,'Position');
fig2 = figure('Units', H.StdUnit,...
'CloseRequestFcn','nncontrolutil(''nnident'',''data_NO_ok'');', ...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'Name', 'Plant Input-Output Data',...
'Tag', 'nnidentdata',...
'NumberTitle', 'off',...
'Position', uipos.fig, ...
'IntegerHandle', 'off',...
'Toolbar', 'none', ...
'Resize', 'off', ...
'WindowStyle', 'modal');
f2.h1=axes('Position',[0.13 0.60 0.74 0.32],'Parent',fig2);
f2.h2=axes('Position',[0.13 0.15 0.74 0.32],'Parent',fig2);
f2.message= uicontrol('Parent',fig2, ...
'Units','points', ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'FontWeight','bold', ...
'ForegroundColor',[0 0 1], ...
'ListboxTop',0, ...
'Position',[156 3 188 20], ...
'Style','text', ...
'Tag','StaticText1');
else
f2=get(fig2,'userdata');
figure(fig2);
end
f2.accept_but = uicontrol('Parent',fig2, ...
'Units','character', ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''data_ok'');', ...
'ListboxTop',0, ...
'Position',[0.5333,0.2051,19,1.5385], ...
'String','Accept Data', ...
'Tag','Pushbutton1');
st=sprintf('The imported data has %d samples.\nPlease Accept or Reject Data to continue.',sam_training);
set(H.error_messages,'string',st);
set(f2.message,'string',st);
else %strcmp(cmd,'gen_data')
a1 = get(H.simulink_file,'string');
udFileEdit = get(H.simulink_file,'UserData');
LastPath = udFileEdit.PathName;
if isempty(LastPath),
a2 = which(cat(2,a1,'.mdl'));
if isempty(a2), a2 = which(cat(2,a1,'.slx')); end
else
a2 = which(cat(2,LastPath,cat(2,a1,'.mdl')));
if isempty(a2), a2 = which(cat(2,LastPath,cat(2,a1,'.slx'))); end
end
if (length(a1) == 0 | length(a2) == 0),
present_error(fig,H,H.simulink_file,a1,0, ...
'You must enter a valid filename for your Simulink plant model');
return
else
sim_file=a1;
OpenFlag=1;
ErrorFlag=isempty(find_system(0,'flat','Name',sim_file));
if ErrorFlag,
ErrorFlag=~(exist(sim_file)==4);
if ~ErrorFlag,
OpenFlag=0;
load_system(sim_file);
end
end
if ErrorFlag,
ErrMsg=[sim_file ' must be the name of a Simulink model.'];
present_error(fig,H,H.simulink_file,a1,0,ErrMsg);
return
end
blk=get_param(sim_file,'blocks');
iblk=0;oblk=0;
for k=1:size(blk,1)
if strcmp(get_param(cat(2,cat(2,sim_file,'/'),blk{k}),'blocktype'),'Inport')
iblk=iblk+1;
end
if strcmp(get_param(cat(2,cat(2,sim_file,'/'),blk{k}),'blocktype'),'Outport')
oblk=oblk+1;
end
end
if ~OpenFlag,close_system(sim_file,0);end
if iblk~=1 | oblk~=1
present_error(fig,H,H.simulink_file,a1,0, ...
'The Simulink plant model must have one Inport and one Outport');
return
end
sim_path=a2(1:strfind(a2,a1)-1); set(H.sim_file_ptr,'userdata',sim_file);
end
a1 = str2num(get(H.max_int_edit,'string'));
max_i_int=get_param(arg1,'max_i_int');
if ~sanitycheckparam(a1) | a1<=0,
present_error(fig,H,H.max_int_edit,max_i_int,1, ...
'You must enter a valid number for the maximum interval value over which the random input is constant.');
return
else max_i_int=a1; set(H.max_i_int_ptr,'userdata',max_i_int); end
a1 = str2num(get(H.min_int_edit,'string'));
min_i_int=get_param(arg1,'min_i_int');
if ~sanitycheckparam(a1) | a1<=0,
present_error(fig,H,H.min_int_edit,min_i_int,1, ...
'You must enter a valid number for the minimum interval value over which the random input is constant.');
return
elseif a1>=max_i_int
present_error(fig,H,H.min_int_edit,min_i_int,1, ...
'You must enter valid maximum and minimum interval values for constant input to the plant.');
return
else min_i_int=a1; set(H.min_i_int_ptr,'userdata',min_i_int); end
a1 = str2num(get(H.Samples,'string'));
sam_training=get_param(arg1,'sam_training');
if ~sanitycheckparam(a1) | (a1 < 1) | (floor(a1)~=a1)
present_error(fig,H,H.Samples,sam_training,1, ...
'You must enter a valid number of samples for training');
return
else sam_training=a1; set(H.sam_training_ptr,'userdata',sam_training); end
Limit_output=get(H.Limit_output_data,'value');
if Limit_output
a1 = str2num(get(H.Max_output,'string'));
max_out=get_param(arg1,'max_output');
if ~sanitycheckparam(a1,false),
present_error(fig,H,H.Max_output,max_out,1, ...
'You must enter a valid maximum plant output');
return
else max_out=a1; set(H.max_out_ptr,'userdata',max_out); end
a1 = str2num(get(H.Min_output,'string'));
min_out=get_param(arg1,'min_output');
if ~sanitycheckparam(a1,false),
present_error(fig,H,H.Min_output,min_out,1, ...
'You must enter a valid minimum plant output');
return
elseif a1>=max_out
present_error(fig,H,H.Min_output,min_out,1, ...
'You must enter valid maximum and minimum plant outputs');
return
else min_out=a1; set(H.min_out_ptr,'userdata',min_out); end
else
max_out=Inf;
set(H.max_out_ptr,'userdata',max_out);
min_out=-Inf;
set(H.min_out_ptr,'userdata',min_out);
end
if isempty(fig2)
pos_fig2=get(fig,'Position');
fig2 = figure('Units', H.StdUnit,...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'CloseRequestFcn','nncontrolutil(''nnident'',''data_NO_ok'');', ...
'Name', 'Plant Input-Output Data',...
'Tag', 'nnidentdata',...
'NumberTitle', 'off',...
'Position', pos_fig2, ...
'IntegerHandle', 'off',...
'Toolbar', 'none', ...
'Resize', 'off', ...
'WindowStyle', 'modal');
f2.h1=axes('Position',[0.13 0.60 0.74 0.32],'Parent',fig2);
f2.h2=axes('Position',[0.13 0.15 0.74 0.32],'Parent',fig2);
f2.message= uicontrol('Parent',fig2, ...
'Units',H.StdUnit, ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'FontWeight','bold', ...
'ForegroundColor',[0 0 1], ...
'ListboxTop',0, ...
'Position',[pos_fig2(3)-52,0.3077,50.1333,2.0513], ...
'Style','text', ...
'Tag','StaticText1');
else
f2=get(fig2,'userdata');
figure(fig2);
end
f2.accept_but = uicontrol('Parent',fig2, ...
'Units','character', ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''stop_sim'');', ...
'ListboxTop',0, ...
'Position',[0.5333,0.2051,19,1.5385], ...
'String','Stop Simulation', ...
'Tag','Pushbutton1');
f2.stop=0;
set(fig2,'UserData',f2);
set(H.error_messages,'string','Simulating plant. Wait until sample data points are generated');
drawnow; % pause needed to refresh the message
options_ini=simset('OutputPoints','all');
options=simset('OutputPoints','all');
step_size=5;
k=1;
k1=1;
% We change cursor shape.
set(fig,'pointer','watch');
Actual_path=pwd;
if isempty(sim_path)
sim_path=Actual_path;
end
cd(sim_path);
tr_dat.Ts=Ts;
min_k=ceil(min_i_int/Ts);
while k<=sam_training %for k=1:sam_training
if ceil((k1-1)/step_size)==(k1-1)/step_size
newsample=rand*(max_i-min_i)+min_i;
k1=1;
step_size=ceil(max([min([(rand*(max_i_int-min_i_int)+min_i_int) max_i_int]) min_i_int])/Ts);
end
k1=k1+1;
tr_dat.U(k,1)=newsample;
% Change to process models with no states
ss=warning;
warning('off');
[time,xx0,yy] = sim(sim_file,[(k-1)*Ts k*Ts],options,[[(k-1)*Ts k*Ts]' [tr_dat.U(k) tr_dat.U(k)]']);
warning(ss);
if size(xx0,1)>0
options.InitialState=xx0(size(xx0,1),:);
end
if k==1
tr_dat.Y(1,1)=yy(1);
end
tr_dat.Y(k+1,1)=yy(size(yy,1));
tr_dat.flag(k,1)=1;
if tr_dat.Y(k+1,1)>max_out | tr_dat.Y(k+1,1)<min_out
options=options_ini;
newsample=rand*(max_i-min_i)+min_i;
k2=1;
while k2<=1 %max_Ni_Nj
k=k+1;
tr_dat.U(k,1)=newsample;
% Change to process models with no states
ss=warning;
warning('off');
[time,xx0,yy] = sim(sim_file,[(k-1)*Ts k*Ts],options,[[(k-1)*Ts k*Ts]' [tr_dat.U(k) tr_dat.U(k)]']);
warning(ss);
if size(xx0,1)>0
options.InitialState=xx0(size(xx0,1),:);
end
tr_dat.Y(k+1,1)=yy(size(yy,1));
tr_dat.flag(k,1)=0;
k2=k2+1;
end
end
% 4-4-00 ODJ. We check for constant output, so we change the input.
% We verified for constant output according to the minimum interval.
% We change the input if we didn't change in for last min_k*2 samples.
% We also reset the k1 and step_size.
if k>(min_k*2+1)
if (abs(mean(tr_dat.Y(k-min_k+1:k+1,1))-mean(tr_dat.Y(k-min_k*2:k-min_k,1)))) < 1e-10 ...
& tr_dat.U(k) == tr_dat.U(k-min_k*2)
newsample=rand*(max_i-min_i)+min_i;
k1=2;
step_size=ceil(max([min([(rand*(max_i_int-min_i_int)+min_i_int) max_i_int]) min_i_int])/Ts);
end
end
if ceil(k/100)==k/100
f2=get(fig2,'userdata');
if f2.stop~=0
st=sprintf('Simulation stopped by the user.\nPlease Accept or Reject Data to continue.');
set(H.error_messages,'string',st);
H.Data_Available=0;
set(fig,'UserData',H);
sam_training=k;
k=k+1;
break
end
st=sprintf('Processing sample # %d of %d total samples.',k,sam_training);
set(H.error_messages,'string',st);
set(f2.message,'string',st);
plot((0:k-1)*Ts,tr_dat.U(1:k),'Parent',f2.h1);
plot((0:k-1)*Ts,tr_dat.Y(2:k+1),'Parent',f2.h2);
set(get(f2.h1,'Title'),'string','Plant Input','fontweight','bold');
set(get(f2.h2,'Title'),'string','Plant Output','fontweight','bold');
set(get(f2.h1,'XLabel'),'string','time (s)');
set(get(f2.h2,'XLabel'),'string','time (s)');
set(fig2,'UserData',f2);
drawnow;
end
k=k+1;
end
if ~f2.stop
st=sprintf('Simulation concluded.\nPlease Accept or Reject Data to continue.');
set(H.error_messages,'string',st);
set(f2.message,'string',st);
end
set(fig,'pointer','arrow');
cd(Actual_path);
tr_dat.U(k,1)=newsample; % We require U and Y have the same size.
set(f2.message,'string',st);
set(f2.accept_but,'Callback','nncontrolutil(''nnident'',''data_ok'');', ...
'String','Accept Data');
end
set(H.Max_input,'enable','off')
set(H.Max_input_text,'enable','off')
set(H.Min_input,'enable','off')
set(H.Min_input_text,'enable','off')
set(H.max_int_edit,'enable','off')
set(H.max_int_text,'enable','off')
set(H.min_int_edit,'enable','off')
set(H.min_int_text,'enable','off')
set(H.Samples_text,'enable','off')
set(H.Samples,'enable','off')
set(H.Sampling_text,'enable','off')
set(H.Sampling_time,'enable','off')
set(H.Max_output,'enable','off')
set(H.Max_output_text,'enable','off')
set(H.Min_output,'enable','off')
set(H.Min_output_text,'enable','off')
set(H.Limit_output_data,'enable','off');
set(H.BrowseButton,'enable','off');
set(H.simulink_file,'enable','off');
set(H.simulink_file_text,'enable','off');
f2.Reject_but = uicontrol('Parent',fig2, ...
'Units','character', ...
'BackgroundColor',[0.752941176470588 0.752941176470588 0.752941176470588], ...
'Callback','nncontrolutil(''nnident'',''data_NO_ok'');', ...
'ListboxTop',0, ...
'Position',[20.5,0.2051,19,1.5385], ...
'String','Reject Data', ...
'Tag','Pushbutton1');
plot((0:sam_training-1)*Ts,tr_dat.U(1:sam_training),'Parent',f2.h1);
plot((0:sam_training-1)*Ts,tr_dat.Y(2:sam_training+1),'Parent',f2.h2);
set(f2.h1,'xlim',[0 (sam_training-1)*Ts]);
set(f2.h2,'xlim',[0 (sam_training-1)*Ts]);
set(get(f2.h1,'Title'),'string','Plant Input','fontweight','bold');
set(get(f2.h2,'Title'),'string','Plant Output','fontweight','bold');
set(get(f2.h1,'XLabel'),'string','time (s)');
set(get(f2.h2,'XLabel'),'string','time (s)');
set(fig,'userdata',H), set(fig2,'UserData',f2);
clear fig fig2 H f2
save(cat(2,tempdir,'nnidentdata2.mat'));
return;
elseif strcmp(cmd,'data_ok')
load(cat(2,tempdir,'nnidentdata2.mat'));
delete(cat(2,tempdir,'nnidentdata2.mat'));
fig2 = findall(0,'type','figure','tag','nnidentdata'); delete(fig2);
N2=length(tr_dat.U);
st=sprintf('Your training data set has %d samples.\nYou can now train the network.',N2-1);
set(H.error_messages,'string',st);
if H.Training_done==1
set(H.Apply_but,'enable','on');
set(H.OK_but,'enable','on');
set(H.Handles.Menus.File.Save_NN,'enable','on')
set(H.Handles.Menus.File.Save_Exit_NN,'enable','on')
end
set(H.Start_but,'enable','on');
set(H.Cancel_but,'enable','on');
H.Data_Available=1;
if H.Data_Imported
H.Data_Generated=0;
set(H.Gen_data_but,'String','Erase Imported Data', ...
'Callback','nncontrolutil(''nnident'',''erase_data'')', ...
'TooltipString','The imported data will be erased and the Training Data section be enabled.');
else
H.Data_Generated=1;
set(H.Gen_data_but,'String','Erase Generated Data', ...
'Callback','nncontrolutil(''nnident'',''erase_data'')', ...
'TooltipString','The generated data will be erased and the Training Data section will be enabled.');
end
set(fig,'userdata',H)
clear fig H
save(cat(2,tempdir,'nnidentdata.mat'));
return
elseif strcmp(cmd,'continue_training')
load(cat(2,tempdir,'nnidentdata.mat'));
HH=msgbox({['The Neural Network is being configured.'] ['Training will start shortly.'] },H.me,'warn');
delete(findobj(HH,'style','pushbutton'));
drawnow; % Replacing pause(1) with drawnow
a1 = str2num(get(H.Hidden_layer_size,'string'));
S1=get_param(arg1,'S1');
if length(a1) == 0,
present_error(fig,H,H.Hidden_layer_size,S1,1, ...
'You must initialize the size of the hidden layer before training the neural network.');
delete(HH);
return
elseif a1<=0 | fix(a1)~=a1 | ~sanitycheckparam(a1),
present_error(fig,H,H.Hidden_layer_size,S1,1, ...
'You must set the size of the hidden layer to a positive integer before generating data or training the neural network');
delete(HH);
return
else S1=a1; set(H.S1_ptr,'userdata',S1); end
a1 = str2num(get(H.Delayed_input,'string'));
if (length(a1) == 0) | (a1 < 1) | (floor(a1)~=a1) | ~sanitycheckparam(a1),
Ni=get_param(arg1,'Ni');
present_error(fig,H,H.Delayed_input,Ni,1, ...
'You must enter a valid number of delayed plant inputs');
delete(HH);
return
else Ni=a1; set(H.Ni_ptr,'userdata',Ni); end
a1 = str2num(get(H.Delayed_output,'string'));
if (length(a1) == 0) | (a1 < 0) | (floor(a1)~=a1) | ~sanitycheckparam(a1),
Nj=get_param(arg1,'Nj');
present_error(fig,H,H.Delayed_output,Nj,1, ...
'You must enter a valid number of delayed plant outputs');
delete(HH);
return
else Nj=a1; set(H.Nj_ptr,'userdata',Nj); end
a1 = get(H.trainfun_edit,'value');
if (a1 < 1) | (a1 > 13),
trainfun=get_param(arg1,'trainfun');
a1=nnstring.first_match(trainfun,func_index);
present_error(fig,H,H.trainfun_edit,a1,0, ...
'Please, correct the training function');
delete(HH);
% set(H.trainfun_edit,'value',a1);
return
else
trainfun=func_index(a1,:);
set_param(arg1,'trainfun',trainfun);
for k1=8:-1:1
if trainfun(k1)==' '
trainfun=trainfun(1:k1-1);
else
break
end
end
end
a1 = str2num(get(H.epochs_h,'string'));
epochs=get_param(arg1,'epochs');
if (length(a1) == 0) | (a1 < 1) | (floor(a1)~=a1) | ~sanitycheckparam(a1),
present_error(fig,H,H.epochs_h,epochs,1, ...
'You must enter a valid number of epochs before training the neural network');
delete(HH);
return
else epochs=a1; set(H.epochs_ptr,'userdata',epochs); end
Use_Previous_Weights=get(H.Use_Previous_Weights_but,'value');
set(H.Use_Previous_Weights_ptr,'userdata',Use_Previous_Weights);
Normalize=get(H.Normalize_data,'Value');
set(H.Normalize_ptr,'userdata',Normalize);
set(H.In_training_ptr,'userdata',1);
% Inputs and targets are preprocessed so that minimum is -1
% and maximum is 1
% if Normalize
[U,up] = mapminmax(tr_dat.U');
U = U'; minp = up.xmin; maxp = up.xmax;
[Y,yp] = mapminmax(tr_dat.Y');
Y = Y'; mint = yp.xmin; maxt = up.xmax;
set(H.minp_ptr,'userdata',minp);
set(H.maxp_ptr,'userdata',maxp);
set(H.mint_ptr,'userdata',mint);
set(H.maxt_ptr,'userdata',maxt);
% else
% ODJ 6/30/99 We need the parameters minp,maxp,mint,maxt for latter simulations.
if Normalize ~= 1
Y=tr_dat.Y;
U=tr_dat.U;
end
Use_Validation=get(H.Use_Validation_but,'value');
Use_Testing=get(H.Use_Testing_but,'value');
N2=length(U);
if Use_Validation & Use_Testing
N1=floor(N2/2);
N3=floor(N2*3/4);
elseif Use_Testing
N1=floor(N2*3/4);
N3=floor(N2*3/4);
elseif Use_Validation
N1=floor(N2*3/4);
N3=N2;
else
N1=N2;
N3=N2;
end
max_Ni_Nj=max([Ni Nj]);
no_valid_data0=find(tr_dat.flag==0);
% 8/27/99 ODJ flag only has information about disrrupted points.
% We must remove as many as delays are necessary to the Neural Network.
no_valid_data=no_valid_data0;
for k=1:max_Ni_Nj-1
no_valid_data=[no_valid_data;no_valid_data0+k];
end
size_no_valid_data=size(no_valid_data,1);
train_points=max_Ni_Nj:N1-1;
valid_points=N1:N3-1;
test_points=N3:N2-1;
for k=1:size_no_valid_data
train_points=train_points(find(train_points(:)~=no_valid_data(k)));
valid_points=valid_points(find(valid_points(:)~=no_valid_data(k)));
test_points=test_points(find(test_points(:)~=no_valid_data(k)));
end
S2=1;
g1 = 'tansig';
g2 = 'purelin';
parent_function=get(H.parent_function_ptr,'userdata');
if strcmp(parent_function,'narma_l2')
ptr=cell(3,1);
ttr=cell(1,1);
vv.P=cell(3,1);
vv.T=cell(1,1);
tt.P=cell(3,1);
tt.T=cell(1,1);
ptr{1,1}=[Y(train_points)'];
vv.P{1,1}=[Y(valid_points)'];
tt.P{1,1}=[Y(test_points)'];
for k=1:Nj-1
ptr{1,1}=[ptr{1,1};[Y(train_points-k)']];
vv.P{1,1}=[vv.P{1,1};Y(valid_points-k)'];
tt.P{1,1}=[tt.P{1,1};Y(test_points-k)'];
end
ptr{3,1}=U(train_points)';
vv.P{3,1}=U(valid_points)';
tt.P{3,1}=U(test_points)';
for k=1:Ni-1
ptr{1,1}=[ptr{1,1};[U(train_points-k)']];
vv.P{1,1}=[vv.P{1,1};U(valid_points-k)'];
tt.P{1,1}=[tt.P{1,1};U(test_points-k)'];
end
ttr{1}=Y(train_points+1)';
vv.T{1}=Y(valid_points+1)';
tt.T{1}=Y(test_points+1)';
ptr{2,1}=ptr{1,1};
vv.P{2,1}=vv.P{1,1};
tt.P{2,1}=tt.P{1,1};
ws = warning('off','NNET:Obsolete');
netn = newff(minmax([ptr{1,1} vv.P{1,1} tt.P{1,1}]),[S1 S2 S1 S2 1 1],{g1,g2,g1,g2,g2,g2},trainfun);
warning(ws)
netn.numInputs=2;
netn.numInputs=3;
netn.inputs{2}.size=netn.inputs{1}.size;
netn.inputs{2}.range=netn.inputs{1}.range;
netn.inputs{3}.range=minmax(ptr{3,1});
netn.biasConnect(5:6)=0;
netn.layers{5}.netInputFcn='netprod';
netn.inputConnect(3,2)=1;
netn.inputConnect(5,3)=1;
netn.layerConnect(6,2)=1;
netn.layerConnect(3,2)=0;
netn=initlay(netn);
IW1_1=netn.IW{1,1};
IW3_2=netn.IW{3,2};
IW5_3=netn.IW{5,3};
LW2_1=netn.LW{2,1};
LW4_3=netn.LW{4,3};
LW5_4=netn.LW{5,4};
LW6_5=netn.LW{6,5};
LW6_2=netn.LW{6,2};
B1=netn.b{1};
B2=netn.b{2};
B3=netn.b{3};
B4=netn.b{4};
if Use_Previous_Weights & ~isempty(nnstring.vert_cat(get_param(arg1,'IW1_1')))
IW1_1b=get(H.IW1_1_ptr,'userdata');
IW3_2b=get(H.IW3_2_ptr,'userdata');
IW5_3b=get(H.IW5_3_ptr,'userdata');
LW2_1b=get(H.LW2_1_ptr,'userdata');
LW4_3b=get(H.LW4_3_ptr,'userdata');
LW5_4b=get(H.LW5_4_ptr,'userdata');
LW6_5b=get(H.LW6_5_ptr,'userdata');
LW6_2b=get(H.LW6_2_ptr,'userdata');
B1b=get(H.B1_ptr,'userdata');
B2b=get(H.B2_ptr,'userdata');
B3b=get(H.B3_ptr,'userdata');
B4b=get(H.B4_ptr,'userdata');
IW1_1a=eval(nnstring.vert_cat(get_param(arg1,'IW1_1')));
IW3_2a=eval(nnstring.vert_cat(get_param(arg1,'IW3_2')));
IW5_3a=eval(nnstring.vert_cat(get_param(arg1,'IW5_3')));
LW2_1a=eval(nnstring.vert_cat(get_param(arg1,'LW2_1')));
LW4_3a=eval(nnstring.vert_cat(get_param(arg1,'LW4_3')));
LW5_4a=eval(nnstring.vert_cat(get_param(arg1,'LW5_4')));
LW6_5a=eval(nnstring.vert_cat(get_param(arg1,'LW6_5')));
LW6_2a=eval(nnstring.vert_cat(get_param(arg1,'LW6_2')));
B1a=eval(nnstring.vert_cat(get_param(arg1,'B1')));
B2a=eval(nnstring.vert_cat(get_param(arg1,'B2')));
B3a=eval(nnstring.vert_cat(get_param(arg1,'B3')));
B4a=eval(nnstring.vert_cat(get_param(arg1,'B4')));
if (size(IW1_1)==size(IW1_1a)) & (size(IW3_2)==size(IW3_2a)) & (size(IW5_3)==size(IW5_3a)) ...
& (size(LW2_1)==size(LW2_1a)) & (size(LW4_3)==size(LW4_3a)) ...
& (size(LW5_4)==size(LW5_4a)) & (size(LW6_5)==size(LW6_5a)) & (size(LW6_2)==size(LW6_2a)) ...
& (size(B1)==size(B1a)) & (size(B2)==size(B2a)) & (size(B3)==size(B3a)) & (size(B4)==size(B4a))
% If Weights different from last generated, we use Simulink weights.
if (size(IW1_1)==size(IW1_1b)) & (size(IW3_2)==size(IW3_2b)) & (size(IW5_3)==size(IW5_3b)) ...
& (size(LW2_1)==size(LW2_1b)) & (size(LW4_3)==size(LW4_3b)) ...
& (size(LW5_4)==size(LW5_4b)) & (size(LW6_5)==size(LW6_5b)) & (size(LW6_2)==size(LW6_2b)) ...
& (size(B1)==size(B1b)) & (size(B2)==size(B2b)) & (size(B3)==size(B3b)) & (size(B4)==size(B4b))
% We only compare IW1_1 to see if we have same values in simulink model and menu.
cx=IW1_1b==IW1_1a;
% Different weights, we ask which we want we prefer.
if sum(cx(:))~=size(IW1_1(:),1)
if ishandle(HH)
delete(HH);
end
switch questdlg(...
{'You have a set of weights in the Simulink model and another set of weights generated in the current training process.'
' ';
'Select which set of weights you want to use. If you select Simulink model weights, the generated weights are discarded.'
' '},...
'Weight Selection','Simulink Model Weights','Generated Weights','Generated Weights');
case 'Simulink Model Weights'
overwriteOK = 1;
case 'Generated Weights'
overwriteOK = 0;
end % switch questdlg
else
overwriteOK = 0;
end
else
overwriteOK = 0;
end
if overwriteOK
netn.IW{1,1}=IW1_1a;
netn.IW{3,2}=IW3_2a;
netn.IW{5,3}=IW5_3a;
netn.LW{2,1}=LW2_1a;
netn.LW{4,3}=LW4_3a;
netn.LW{5,4}=LW5_4a;
netn.LW{6,5}=LW6_5a;
netn.LW{6,2}=LW6_2a;
netn.b{1}=B1a;
netn.b{2}=B2a;
netn.b{3}=B3a;
netn.b{4}=B4a;
else
netn.IW{1,1}=IW1_1b;
netn.IW{3,2}=IW3_2b;
netn.IW{5,3}=IW5_3b;
netn.LW{2,1}=LW2_1b;
netn.LW{4,3}=LW4_3b;
netn.LW{5,4}=LW5_4b;
netn.LW{6,5}=LW6_5b;
netn.LW{6,2}=LW6_2b;
netn.b{1}=B1b;
netn.b{2}=B2b;
netn.b{3}=B3b;
netn.b{4}=B4b;
end
end
end
else
ptr=[];
vv.P=[];
tt.P=[];
for k=1:Nj
ptr=[ptr;[Y(train_points-k+1)']];
vv.P=[vv.P;Y(valid_points-k+1)'];
tt.P=[tt.P;Y(test_points-k+1)'];
end
for k=1:Ni
ptr=[ptr;[U(train_points+1-k)']];
vv.P=[vv.P;U(valid_points-k+1)'];
tt.P=[tt.P;U(test_points-k+1)'];
end
ttr=Y(train_points+1)';
vv.T=Y(valid_points+1)';
tt.T=Y(test_points+1)';
ws = warning('off','NNET:Obsolete');
netn = newff(minmax([ptr vv.P tt.P]),[S1 S2],{g1,g2},trainfun);
warning(ws)
netn2=netn;
inputsrange=netn2.inputs{1}.range;
iw=netn2.IW;
netn2.inputs{1}.size=1;
netn2.inputs{1}.range=inputsrange(Nj+1,:);
netn2.layerconnect(1,2)=1;
netn2.inputweights{1}.delays=[1:Ni];
netn2.layerweights{1,2}.delays=[1:Nj];
netn2.IW{1}=iw{1}(:,Nj+1:Ni+Nj);
netn2.LW{1,2}=iw{1}(:,1:Nj);
IW=netn2.IW{1,1};
LW1_2=netn2.LW{1,2};
LW2_1=netn2.LW{2,1};
B1=netn2.b{1};
B2=netn2.b{2};
if Use_Previous_Weights & ~isempty(nnstring.vert_cat(get_param(arg1,'IW')))
IWb=get(H.IW_ptr,'userdata');
LW2_1b=get(H.LW2_1_ptr,'userdata');
LW1_2b=get(H.LW1_2_ptr,'userdata');
B1b=get(H.B1_ptr,'userdata');
B2b=get(H.B2_ptr,'userdata');
IW_2=eval(nnstring.vert_cat(get_param(arg1,'IW')));
LW2_1_2=eval(nnstring.vert_cat(get_param(arg1,'LW2_1')));
LW1_2_2=eval(nnstring.vert_cat(get_param(arg1,'LW1_2')));
B1_2=eval(nnstring.vert_cat(get_param(arg1,'B1')));
B2_2=eval(nnstring.vert_cat(get_param(arg1,'B2')));
if (size(IW_2)==size(IW)) & (size(LW2_1_2)==size(LW2_1)) & (size(LW1_2_2)==size(LW1_2)) ...
& (size(B1_2)==size(B1)) & (size(B2_2)==size(B2))
% If Weights different from last generated, we use Simulink weights.
if (size(IWb)==size(IW)) & (size(LW2_1b)==size(LW2_1)) & (size(LW1_2b)==size(LW1_2)) ...
& (size(B1b)==size(B1)) & (size(B2b)==size(B2))
% We only compare IW1_1 to see if we have same values in simulink model and menu.
cx=IWb==IW_2;
% Different weights, we ask which we want we prefer.
if sum(cx(:))~=size(IW_2(:),1)
if ishandle(HH)
delete(HH);
end
switch questdlg(...
{'You have a set of weights in the Simulink model and another set of weights generated in the current training process.'
' ';
'Select which set of weights you want to use. If you select Simulink model weights, the generated weights are discarded.'
' '},...
'Weight Selection','Simulink Model Weights','Generated Weights','Generated Weights');
case 'Simulink Model Weights'
overwriteOK = 1;
case 'Generated Weights'
overwriteOK = 0;
end % switch questdlg
else
overwriteOK = 0;
end
else
overwriteOK = 0;
end
if overwriteOK
netn.b{1}=B1_2;
netn.b{2}=B2_2;
netn.LW{2,1}=LW2_1_2;
netn.IW{1,1}=[LW1_2_2 IW_2];
else
netn.b{1}=B1b;
netn.b{2}=B2b;
netn.LW{2,1}=LW2_1b;
netn.IW{1,1}=[LW1_2b IWb];
end
end
end
end
% Training function could be changed in continue training.
a1 = get(H.trainfun_edit,'value');
if (a1 < 1) | (a1 > 13),
trainfun=get_param(arg1,'trainfun');
a1=nnstring.first_match(trainfun,func_index);
present_error(fig,H,H.trainfun_edit,a1,0, ...
'Please, correct the training function');
if ishandle(HH)
delete(HH);
end
return
else
trainfun=func_index(a1,:);
set_param(arg1,'trainfun',trainfun);
for k1=8:-1:1
if trainfun(k1)==' '
trainfun=trainfun(1:k1-1);
else
break
end
end
end
netn.trainFcn=trainfun;
end
netn.trainParam.epochs = epochs;
netn.trainParam.show = 1;
netn.trainParam.min_grad=1e-10;
set(H.error_messages,'string','Training Neural Network');
set(fig,'pointer','watch');
if ishandle(HH)
delete(HH);
end
drawnow; % Replacing pause(1) with drawnow
if ~Use_Testing & ~Use_Validation
[netn,tr] = train(netn,ptr,ttr);
elseif ~Use_Testing
[netn,tr] = train(netn,ptr,ttr,[],[],vv);
elseif ~Use_Validation
[netn,tr] = train(netn,ptr,ttr,[],[],[],tt);
else
[netn,tr] = train(netn,ptr,ttr,[],[],vv,tt);
end
set(H.Simulating_text,'visible','off'); clear fig H
save(cat(2,tempdir,'nnidentdata.mat'));
fig = findall(0,'type','figure','tag','nnident'); H = get(fig,'userdata');
parent_function=get(H.parent_function_ptr,'userdata');
if strcmp(parent_function,'nnpredict')
title_fig2='NN Predictive Control';
elseif strcmp(parent_function,'nnmodref')
title_fig2='NN Model Reference Control';
elseif strcmp(parent_function,'narma_l2')
title_fig2='NN NARMA L2';
end
Ysim = netn(ptr);
fig2 = findall(0,'type','figure','tag',cat(2,parent_function,'_train'));
matlab_position=get(0,'screensize');
matlab_units=get(0,'units');
if strcmp(matlab_units,'pixels');
matlab_position=matlab_position*H.PointsToPixels;
end
if isempty(fig2)
fig2_position=[max([30 matlab_position(3)-410]) 90 328 335];
fig2 = figure('Units', 'points',...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'Name', cat(2,'Training data for ',title_fig2),...
'Tag', cat(2,parent_function,'_train'),...
'NumberTitle', 'off',...
'Position', fig2_position, ...
'IntegerHandle', 'off',...
'Resize', 'off', ...
'Toolbar', 'none');
f2.h1=axes('Position',[0.13 0.58 0.32 0.34],'Parent',fig2); %subplot(221);
f2.h2=axes('Position',[0.57 0.58 0.32 0.34],'Parent',fig2); %subplot(222);
f2.h3=axes('Position',[0.13 0.11 0.32 0.34],'Parent',fig2); %subplot(223);
f2.h4=axes('Position',[0.57 0.11 0.32 0.34],'Parent',fig2); %subplot(224);
else
f2=get(fig2,'userdata');
figure(fig2);
end
if strcmp(parent_function,'narma_l2')
if Normalize
plot((0:size(ptr{1,1},2)-1)*Ts,(ptr{3,1}+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(ptr{1,1},2)-1)*Ts,(ttr{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([ttr{1} Ysim{1} -1]) max([ttr{1} Ysim{1} 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(ptr{1,1},2)-1)*Ts,(ttr{1}-Ysim{1})*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(ptr{1,1},2)-1)*Ts,(Ysim{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([ttr{1} Ysim{1} -1]) max([ttr{1} Ysim{1} 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(ptr{1,1},2)-1)*Ts,ptr{3,1},'Parent',f2.h1);
plot((0:size(ptr{1,1},2)-1)*Ts,ttr{1},'Parent',f2.h2);
set(f2.h2,'ylim',[min([ttr{1} Ysim{1} mint]) max([ttr{1} Ysim{1} maxt])]);
plot((0:size(ptr{1,1},2)-1)*Ts,ttr{1}-Ysim{1},'Parent',f2.h3);
plot((0:size(ptr{1,1},2)-1)*Ts,Ysim{1},'Parent',f2.h4);
set(f2.h4,'ylim',[min([ttr{1} Ysim{1} mint]) max([ttr{1} Ysim{1} maxt])]);
end
set(f2.h1,'xlim',[0 (size(ptr{1,1},2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(ptr{1,1},2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(ptr{1,1},2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(ptr{1,1},2)-1)*Ts]);
else
if Normalize
plot((0:size(ptr,2)-1)*Ts,(ptr(Nj+1,:)+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(ptr,2)-1)*Ts,(ttr+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([ttr Ysim -1]) max([ttr Ysim 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(ptr,2)-1)*Ts,(ttr-Ysim)*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(ptr,2)-1)*Ts,(Ysim+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([ttr Ysim -1]) max([ttr Ysim 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(ptr,2)-1)*Ts,ptr(Nj+1,:),'Parent',f2.h1);
plot((0:size(ptr,2)-1)*Ts,ttr,'Parent',f2.h2);
set(f2.h2,'ylim',[min([ttr Ysim mint]) max([ttr Ysim maxt])]);
plot((0:size(ptr,2)-1)*Ts,ttr-Ysim,'Parent',f2.h3);
plot((0:size(ptr,2)-1)*Ts,Ysim,'Parent',f2.h4);
set(f2.h4,'ylim',[min([ttr Ysim mint]) max([ttr Ysim maxt])]);
end
set(f2.h1,'xlim',[0 (size(ptr,2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(ptr,2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(ptr,2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(ptr,2)-1)*Ts]);
end
set(get(f2.h1,'Title'),'string','Input','fontweight','bold');
set(get(f2.h2,'Title'),'string','Plant Output','fontweight','bold');
set(get(f2.h3,'Title'),'string','Error','fontweight','bold');
set(get(f2.h4,'Title'),'string','NN Output','fontweight','bold');
set(get(f2.h3,'XLabel'),'string','time (s)');
set(get(f2.h4,'XLabel'),'string','time (s)');
set(fig2,'UserData',f2);
if Use_Validation
Yvsim = netn(vv.P);
fig2=findall(0,'type','figure','tag',cat(2,parent_function,'_valid'));
if isempty(fig2)
fig2_position=[max([60 matlab_position(3)-380]) 60 328 335];
fig2 = figure('Units', 'points',...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'Name', cat(2,'Validation data for ',title_fig2),...
'Tag', cat(2,parent_function,'_valid'),...
'NumberTitle', 'off',...
'Position', fig2_position, ...
'IntegerHandle', 'off',...
'Resize', 'off', ...
'Toolbar', 'none');
f2.h1=axes('Position',[0.13 0.58 0.32 0.34],'Parent',fig2); %subplot(221);
f2.h2=axes('Position',[0.57 0.58 0.32 0.34],'Parent',fig2); %subplot(222);
f2.h3=axes('Position',[0.13 0.11 0.32 0.34],'Parent',fig2); %subplot(223);
f2.h4=axes('Position',[0.57 0.11 0.32 0.34],'Parent',fig2); %subplot(224);
else
f2=get(fig2,'userdata');
figure(fig2);
end
if strcmp(parent_function,'narma_l2')
if Normalize
plot((0:size(vv.P{1,1},2)-1)*Ts,(vv.P{3,1}+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(vv.P{1,1},2)-1)*Ts,(vv.T{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([vv.T{1} Yvsim{1} -1]) max([vv.T{1} Yvsim{1} 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(vv.P{1,1},2)-1)*Ts,(vv.T{1}-Yvsim{1})*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(vv.P{1,1},2)-1)*Ts,(Yvsim{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([vv.T{1} Yvsim{1} -1]) max([vv.T{1} Yvsim{1} 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(vv.P{1,1},2)-1)*Ts,vv.P{3,1},'Parent',f2.h1);
plot((0:size(vv.P{1,1},2)-1)*Ts,vv.T{1},'Parent',f2.h2);
set(f2.h2,'ylim',[min([vv.T{1} Yvsim{1} mint]) max([vv.T{1} Yvsim{1} maxt])]);
plot((0:size(vv.P{1,1},2)-1)*Ts,vv.T{1}-Yvsim{1},'Parent',f2.h3);
plot((0:size(vv.P{1,1},2)-1)*Ts,Yvsim{1},'Parent',f2.h4);
set(f2.h4,'ylim',[min([vv.T{1} Yvsim{1} mint]) max([vv.T{1} Yvsim{1} maxt])]);
end
set(f2.h1,'xlim',[0 (size(vv.P{1,1},2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(vv.P{1,1},2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(vv.P{1,1},2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(vv.P{1,1},2)-1)*Ts]);
else
if Normalize
plot((0:size(vv.P,2)-1)*Ts,(vv.P(Nj+1,:)+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(vv.P,2)-1)*Ts,(vv.T+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([vv.T Yvsim -1]) max([vv.T Yvsim 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(vv.P,2)-1)*Ts,(vv.T-Yvsim)*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(vv.P,2)-1)*Ts,(Yvsim+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([vv.T Yvsim -1]) max([vv.T Yvsim 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(vv.P,2)-1)*Ts,vv.P(Nj+1,:),'Parent',f2.h1);
plot((0:size(vv.P,2)-1)*Ts,vv.T,'Parent',f2.h2);
set(f2.h2,'ylim',[min([vv.T Yvsim mint]) max([vv.T Yvsim maxt])]);
plot((0:size(vv.P,2)-1)*Ts,vv.T-Yvsim,'Parent',f2.h3);
plot((0:size(vv.P,2)-1)*Ts,Yvsim,'Parent',f2.h4);
set(f2.h4,'ylim',[min([vv.T Yvsim mint]) max([vv.T Yvsim maxt])]);
end
set(f2.h1,'xlim',[0 (size(vv.P,2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(vv.P,2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(vv.P,2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(vv.P,2)-1)*Ts]);
end
set(get(f2.h1,'Title'),'string','Input','fontweight','bold');
set(get(f2.h2,'Title'),'string','Plant Output','fontweight','bold');
set(get(f2.h3,'Title'),'string','Error','fontweight','bold');
set(get(f2.h4,'Title'),'string','NN Output','fontweight','bold');
set(get(f2.h3,'XLabel'),'string','time (s)');
set(get(f2.h4,'XLabel'),'string','time (s)');
set(fig2,'UserData',f2);
end
if Use_Testing
[Ytsim,Pf,Af] = netn(tt.P);
fig2=findall(0,'type','figure','tag',cat(2,parent_function,'_test'));
if isempty(fig2)
fig2_position=[max([90 matlab_position(3)-350]) 30 328 335];
fig2 = figure('Units', 'points',...
'Interruptible','off', ...
'BusyAction','cancel', ...
'HandleVis','Callback', ...
'Name', cat(2,'Testing data for ',title_fig2),...
'Tag', cat(2,parent_function,'_test'),...
'NumberTitle', 'off',...
'Position', fig2_position, ...
'IntegerHandle', 'off',...
'Resize', 'off', ...
'Toolbar', 'none');
f2.h1=axes('Position',[0.13 0.58 0.32 0.34],'Parent',fig2); %subplot(221);
f2.h2=axes('Position',[0.57 0.58 0.32 0.34],'Parent',fig2); %subplot(222);
f2.h3=axes('Position',[0.13 0.11 0.32 0.34],'Parent',fig2); %subplot(223);
f2.h4=axes('Position',[0.57 0.11 0.32 0.34],'Parent',fig2); %subplot(224);
else
f2=get(fig2,'userdata');
figure(fig2);
end
if strcmp(parent_function,'narma_l2')
if Normalize
plot((0:size(tt.P{1,1},2)-1)*Ts,(tt.P{3,1}+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(tt.P{1,1},2)-1)*Ts,(tt.T{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([tt.T{1} Ytsim{1} -1]) max([tt.T{1} Ytsim{1} 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(tt.P{1,1},2)-1)*Ts,(tt.T{1}-Ytsim{1})*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(tt.P{1,1},2)-1)*Ts,(Ytsim{1}+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([tt.T{1} Ytsim{1} -1]) max([tt.T{1} Ytsim{1} 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(tt.P{1,1},2)-1)*Ts,tt.P{3,1},'Parent',f2.h1);
plot((0:size(tt.P{1,1},2)-1)*Ts,tt.T{1},'Parent',f2.h2);
set(f2.h2,'ylim',[min([tt.T{1} Ytsim{1} mint]) max([tt.T{1} Ytsim{1} maxt])]);
plot((0:size(tt.P{1,1},2)-1)*Ts,tt.T{1}-Ytsim{1},'Parent',f2.h3);
plot((0:size(tt.P{1,1},2)-1)*Ts,Ytsim{1},'Parent',f2.h4);
set(f2.h4,'ylim',[min([tt.T{1} Ytsim{1} mint]) max([tt.T{1} Ytsim{1} maxt])]);
end
set(f2.h1,'xlim',[0 (size(tt.P{1,1},2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(tt.P{1,1},2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(tt.P{1,1},2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(tt.P{1,1},2)-1)*Ts]);
else
if Normalize
plot((0:size(tt.P,2)-1)*Ts,(tt.P(Nj+1,:)+1)*(maxp-minp)/2+minp,'Parent',f2.h1);
plot((0:size(tt.P,2)-1)*Ts,(tt.T+1)*(maxt-mint)/2+mint,'Parent',f2.h2);
set(f2.h2,'ylim',([min([tt.T Ytsim -1]) max([tt.T Ytsim 1])]+1)*abs(maxt-mint)/2+mint);
plot((0:size(tt.P,2)-1)*Ts,(tt.T-Ytsim)*(maxt-mint)/2,'Parent',f2.h3);
plot((0:size(tt.P,2)-1)*Ts,(Ytsim+1)*(maxt-mint)/2+mint,'Parent',f2.h4);
set(f2.h4,'ylim',([min([tt.T Ytsim -1]) max([tt.T Ytsim 1])]+1)*abs(maxt-mint)/2+mint);
else
plot((0:size(tt.P,2)-1)*Ts,tt.P(Nj+1,:),'Parent',f2.h1);
plot((0:size(tt.P,2)-1)*Ts,tt.T,'Parent',f2.h2);
set(f2.h2,'ylim',[min([tt.T Ytsim mint]) max([tt.T Ytsim maxt])]);
plot((0:size(tt.P,2)-1)*Ts,tt.T-Ytsim,'Parent',f2.h3);
plot((0:size(tt.P,2)-1)*Ts,Ytsim,'Parent',f2.h4);
set(f2.h4,'ylim',[min([tt.T Ytsim mint]) max([tt.T Ytsim maxt])]);
end
set(f2.h1,'xlim',[0 (size(tt.P,2)-1)*Ts]);
set(f2.h2,'xlim',[0 (size(tt.P,2)-1)*Ts]);
set(f2.h3,'xlim',[0 (size(tt.P,2)-1)*Ts]);
set(f2.h4,'xlim',[0 (size(tt.P,2)-1)*Ts]);
end
set(get(f2.h1,'Title'),'string','Input','fontweight','bold');
set(get(f2.h2,'Title'),'string','Plant Output','fontweight','bold');
set(get(f2.h3,'Title'),'string','Error','fontweight','bold');
set(get(f2.h4,'Title'),'string','NN Output','fontweight','bold');
set(get(f2.h3,'XLabel'),'string','time (s)');
set(get(f2.h4,'XLabel'),'string','time (s)');
set(fig2,'UserData',f2);
end
if strcmp(parent_function,'narma_l2')
IW1_1=netn.IW{1,1};
IW3_2=netn.IW{3,2};
IW5_3=netn.IW{5,3};
LW2_1=netn.LW{2,1};
LW4_3=netn.LW{4,3};
LW5_4=netn.LW{5,4};
LW6_5=netn.LW{6,5};
LW6_2=netn.LW{6,2};
B1=netn.b{1};
B2=netn.b{2};
B3=netn.b{3};
B4=netn.b{4};
set(H.IW1_1_ptr,'userdata',IW1_1);
set(H.IW3_2_ptr,'userdata',IW3_2);
set(H.IW5_3_ptr,'userdata',IW5_3);
set(H.LW2_1_ptr,'userdata',LW2_1);
set(H.LW4_3_ptr,'userdata',LW4_3);
set(H.LW5_4_ptr,'userdata',LW5_4);
set(H.LW6_5_ptr,'userdata',LW6_5);
set(H.LW6_2_ptr,'userdata',LW6_2);
set(H.B1_ptr,'userdata',B1);
set(H.B2_ptr,'userdata',B2);
set(H.B3_ptr,'userdata',B3);
set(H.B4_ptr,'userdata',B4);
else
netn2=netn;
inputsrange=netn2.inputs{1}.range;
iw=netn2.IW;
netn2.inputs{1}.size=1;
netn2.inputs{1}.range=inputsrange(Nj+1,:);
netn2.layerconnect(1,2)=1;
netn2.inputweights{1}.delays=[1:Ni];
netn2.layerweights{1,2}.delays=[1:Nj];
netn2.IW{1}=iw{1}(:,Nj+1:Ni+Nj);
netn2.LW{1,2}=iw{1}(:,1:Nj);
netn2.layerweights{1,2}.learnFcn='learngdm';
IW=netn2.IW{1,1};
LW1_2=netn2.LW{1,2};
LW2_1=netn2.LW{2,1};
B1=netn2.b{1};
B2=netn2.b{2};
set(H.IW_ptr,'userdata',IW);
set(H.LW1_2_ptr,'userdata',LW1_2);
set(H.LW2_1_ptr,'userdata',LW2_1);
set(H.B1_ptr,'userdata',B1);
set(H.B2_ptr,'userdata',B2);
end
set(H.error_messages,'string','Training complete. You can generate or import new data, continue training or save results by selecting OK or Apply.');
H.Training_done=1;
set(H.Apply_but,'enable','on');
set(H.OK_but,'enable','on');
set(H.Handles.Menus.File.Save_NN,'enable','on')
set(H.Handles.Menus.File.Save_Exit_NN,'enable','on')
set(H.Start_but,'enable','on');
set(H.Cancel_but,'enable','on');
set(H.Use_Previous_Weights_ptr,'userdata',1);
set(H.Use_Previous_Weights_but,'value',1);
set(H.In_training_ptr,'userdata',0);
set(fig,'userdata',H,'pointer','arrow');
figure(fig);
elseif strcmp(cmd,'stop_sim')
fig2=findall(0,'type','figure','tag','nnidentdata');
f2=get(fig2,'userdata');
f2.stop=1;
set(fig2,'UserData',f2);
return;
elseif (strcmp(cmd,'apply') | strcmp(cmd,'ok')) && ~isempty(fig)
if get(H.In_training_ptr,'userdata')~=0
return
end
arg1=get(H.gcbh_ptr,'userdata');
S1=get(H.S1_ptr,'userdata');
sim_file=get(H.sim_file_ptr,'userdata');
Ts=get(H.Ts_ptr,'userdata');
Ni=get(H.Ni_ptr,'userdata');
Nj=get(H.Nj_ptr,'userdata');
Use_Previous_Weights = get(H.Use_Previous_Weights_ptr,'userdata');
Use_Validation=get(H.Use_Validation_but,'value');
Use_Testing=get(H.Use_Testing_but,'value');
max_i=get(H.max_i_ptr,'userdata');
min_i=get(H.min_i_ptr,'userdata');
max_i_int=get(H.max_i_int_ptr,'userdata');
min_i_int=get(H.min_i_int_ptr,'userdata');
sam_training=get(H.sam_training_ptr,'userdata');
epochs=get(H.epochs_ptr,'userdata');
parent_function=get(H.parent_function_ptr,'userdata');
if strcmp(parent_function,'narma_l2')
IW1_1=get(H.IW1_1_ptr,'userdata');
IW3_2=get(H.IW3_2_ptr,'userdata');
IW5_3=get(H.IW5_3_ptr,'userdata');
LW2_1=get(H.LW2_1_ptr,'userdata');
LW4_3=get(H.LW4_3_ptr,'userdata');
LW5_4=get(H.LW5_4_ptr,'userdata');
LW6_5=get(H.LW6_5_ptr,'userdata');
LW6_2=get(H.LW6_2_ptr,'userdata');
B1=get(H.B1_ptr,'userdata');
B2=get(H.B2_ptr,'userdata');
B3=get(H.B3_ptr,'userdata');
B4=get(H.B4_ptr,'userdata');
set_param(arg1,'IW1_1',mat2str(IW1_1,20));
set_param(arg1,'IW3_2',mat2str(IW3_2,20));
set_param(arg1,'IW5_3',mat2str(IW5_3,20));
set_param(arg1,'LW2_1',mat2str(LW2_1,20));
set_param(arg1,'LW4_3',mat2str(LW4_3,20));
set_param(arg1,'LW5_4',mat2str(LW5_4,20));
set_param(arg1,'LW6_5',mat2str(LW6_5,20));
set_param(arg1,'LW6_2',mat2str(LW6_2,20));
set_param(arg1,'B1',mat2str(B1,20));
set_param(arg1,'B2',mat2str(B2,20));
set_param(arg1,'B3',mat2str(B3,20));
set_param(arg1,'B4',mat2str(B4,20));
else
IW=get(H.IW_ptr,'userdata');
LW2_1=get(H.LW2_1_ptr,'userdata');
LW1_2=get(H.LW1_2_ptr,'userdata');
B1=get(H.B1_ptr,'userdata');
B2=get(H.B2_ptr,'userdata');
set_param(arg1,'IW',mat2str(IW,20));
set_param(arg1,'LW1_2',mat2str(LW1_2,20));
set_param(arg1,'LW2_1',mat2str(LW2_1,20));
set_param(arg1,'B1',mat2str(B1,20));
set_param(arg1,'B2',mat2str(B2,20));
end
minp=get(H.minp_ptr,'userdata');
maxp=get(H.maxp_ptr,'userdata');
mint=get(H.mint_ptr,'userdata');
maxt=get(H.maxt_ptr,'userdata');
Normalize=get(H.Normalize_ptr,'userdata');
Limit_output=get(H.Limit_output_data,'value');
max_out=get(H.max_out_ptr,'userdata');
min_out=get(H.min_out_ptr,'userdata');
set_param(arg1,'S1',num2str(S1));
set_param(arg1,'sim_file',sim_file);
set_param(arg1,'Ts',num2str(Ts));
set_param(arg1,'Ni',num2str(Ni));
set_param(arg1,'Nj',num2str(Nj));
set_param(arg1,'Use_Previous_Weights',num2str(Use_Previous_Weights));
set_param(arg1,'Use_Validation',num2str(Use_Validation));
set_param(arg1,'Use_Testing',num2str(Use_Testing));
set_param(arg1,'max_i',num2str(max_i));
set_param(arg1,'min_i',num2str(min_i));
set_param(arg1,'max_i_int',num2str(max_i_int));
set_param(arg1,'min_i_int',num2str(min_i_int));
set_param(arg1,'sam_training',num2str(sam_training));
set_param(arg1,'epochs',num2str(epochs));
set_param(arg1,'minp',num2str(minp,20));
set_param(arg1,'maxp',num2str(maxp,20));
set_param(arg1,'mint',num2str(mint,20));
set_param(arg1,'maxt',num2str(maxt,20));
set_param(arg1,'Normalize',num2str(Normalize));
set_param(arg1,'limit_output',num2str(Limit_output));
set_param(arg1,'max_output',num2str(max_out));
set_param(arg1,'min_output',num2str(min_out));
if strcmp(cmd,'ok')
if ~strcmp(parent_function,'narma_l2')
arg2=get(H.gcb_ptr,'userdata');
feval(parent_function,'',arg1,arg2,'nnident');
end
delete(fig)
if exist(cat(2,tempdir,'nnidentdata.mat'))
delete(cat(2,tempdir,'nnidentdata.mat'));
end
end
elseif strcmp(cmd,'check_params')
if nargin < 4 % arg3 indicates whether to check for infs or not
arg3 = true;
end
checkparam(arg1, H, arg2, arg3);
end
function present_error(fig,H,text_field,field_value,field_type,theMessage)
if H.Data_Available
set(H.Start_but,'enable','on')
end
if H.Training_done
set(H.OK_but,'enable','on')
set(H.Apply_but,'enable','on')
set(H.Handles.Menus.File.Save_NN,'enable','on')
set(H.Handles.Menus.File.Save_Exit_NN,'enable','on')
end
set(H.Cancel_but,'enable','on');
if text_field~=0
if field_type % Number
set(text_field,'string',num2str(field_value));
else % ASCII or No change.
set(text_field,'string',field_value);
end
else
text_field=0;
end
set(H.error_messages,'string',theMessage);
errordlg(theMessage,'Plant Identification Warning','modal');
set(fig,'pointer','arrow');
function paramok = checkparam(param2check, handles, paramlabel, checkforinf)
paramok = true; %set to true initially
paramH = getfield(handles, param2check);
paramval = str2num(get(paramH, 'String'));
try
% Common Checks for all params
theMessage = 'Illegal value assigned to parameter';
if ~sanitycheckparam(paramval, checkforinf)
nnerr.throw('Parameters',theMessage);
end
catch
theMessage = sprintf('Illegal value assigned to ''%s'' parameter', paramlabel);
errordlg(theMessage,'Plant Identification Warning','modal');
paramok = false;
end
function paramok = sanitycheckparam(param, checkforinf)
if nargin < 2
checkforinf = true;
end
if iscell(param) || ~isscalar(param) ...
|| isempty(param) || ~isnumeric(param) ...
|| isnan(param) || ~isreal(param)
paramok = false;
return;
end
if checkforinf
if isinf(param)
paramok = false;
return;
end
end
paramok = true;
function uipos = getuipos
tlabelw = 25;
labelw = 32;
editw = 12;
border = 1.3333;
labelh = 1.5;
edith = 1.53846;
figw = 2* ((border*2) + labelw + editw + (2*border)); %97.3333;
figh = 36.7179;
butwbig = (figw-(border*10))/3;
butwsmall = (figw-(border*11)-butwbig)/3;
sunits = get(0, 'Units');
set (0, 'Units', 'character'); set (0, 'Units', 'points');
ssinchar = get(0, 'ScreenSize'); ssinchar = ssinchar ./ [1 1 3.75 9.75];
set (0, 'Units', sunits);
figl = (ssinchar(3) - figw) / 2;
figb = (ssinchar(4) - figh) / 2;
uipos.fig = [figl,figb,figw,figh];
framew = figw - (border*2);
uipos.frame4 = [border,0.205128,framew,2.25641];
uipos.frame5 = [border,2.66667,framew,6.87179];
uipos.frame1 = [border,10.5641,framew,14.0513];
uipos.frame6= [border,25.4359,framew,7.48718];
labell = (figw-tlabelw)/2;
uipos.h1_1 = [labell,8.5641,tlabelw,labelh];
uipos.h1_2 = [labell,23.5641,tlabelw,labelh];
uipos.h1_3 = [labell,31.8718,tlabelw,labelh];
uipos.Hidden_layer_text = [border*2,30.159,labelw,labelh];
uipos.Sampling_text = [border*2,28.0051,labelw,labelh];
uipos.Samples_text = [border*2,22.0564,labelw,labelh];
uipos.Max_input_text = [border*2,19.8,labelw,labelh];
uipos.Min_input_text = [border*2,17.5436,labelw,labelh];
uipos.max_int_text = [border*2,15.2872,labelw,labelh];
uipos.min_int_text = [border*2,13.0308,labelw,labelh];
uipos.epochs_text = [border*2,6.46667,labelw,labelh];
uipos.Normalize_data = [border*4,25.7487,labelw,labelh];
fc2l = border*2 + labelw + 1.5;
uipos.Hidden_layer_size = [fc2l,30.359,editw,edith];
uipos.Sampling_time = [fc2l,28.2051,editw,edith];
uipos.Samples = [fc2l,22.2564,editw,edith];
uipos.Max_input = [fc2l,20,editw,edith];
uipos.Min_input = [fc2l,17.7436,editw,edith];
uipos.max_int_edit = [fc2l,15.4872,editw,edith];
uipos.min_int_edit = [fc2l,13.2308,editw,edith];
uipos.epochs_h = [fc2l,6.66667,editw,edith];
scl = fc2l + editw + (border*2);
uipos.Delayed_input_text = [scl,30.159,labelw,labelh];
uipos.Delayed_output_text = [scl,27.9026,labelw,labelh];
uipos.Max_output_text = [scl,19.8,labelw,labelh];
uipos.Min_output_text = [scl,17.5436,labelw,labelh];
uipos.simulink_file_text = [scl,15.2872,labelw,labelh];
sc2l = scl + labelw + border;
uipos.Delayed_input = [sc2l,30.359,editw,edith];
uipos.Delayed_output = [sc2l,28.1026,editw,edith];
uipos.Max_output = [sc2l,20,editw,edith];
uipos.Min_output = [sc2l,17.7436,editw,edith];
uipos.BrowseButton = [sc2l,15.4872,editw,1.64103];
uipos.Limit_output_data = [sc2l+editw-labelw-3,22.2564,labelw,labelh];
uipos.trainfun_text = [scl,6.46667,labelw-(0.5*editw),edith];
uipos.trainfun_edit = [sc2l-(0.5*editw),6.66667,editw*1.5,edith];
uipos.simulink_file = [scl,13.2308,labelw+editw+border,labelh];
uipos.Title_nnident = [(figw-framew)/2,33.8462,framew,2.23077];
uipos.Gen_data_but = [border*4,10.9487,butwbig,1.69231];
uipos.Get_data_file_but = [(figw-butwbig)/2,10.9487,butwbig,1.69231];
uipos.Save_to_file_but = [figw-(border*4)-butwbig,10.9487,butwbig,1.69231];
uipos.Use_Previous_Weights_but = [border*4,4.92308,butwbig,edith];
uipos.Use_Validation_but = [(figw-butwbig)/2,4.92308,butwbig,edith];
uipos.Use_Testing_but = [figw-(border*4)-butwbig,4.92308,butwbig,edith];
uipos.Start_but = [border*4,3.07692,butwbig,1.64103];
uipos.OK_but = [uipos.Start_but(1) + border + butwbig,3.07692,butwsmall,1.64103];
uipos.Cancel_but = [uipos.OK_but(1) + border + butwsmall,3.07692,butwsmall,1.64103];
uipos.Apply_but = [uipos.Cancel_but(1) + border + butwsmall,3.07692,butwsmall,1.64103];
uipos.Simulating_text = [2.4,5.12821,32,labelh];
uipos.error_messages = [border+(0.3*border),0.307692,framew-(0.6*border),2.05128];
function x = convertEmptyBrackets(x)
if ischar(x) && strcmp(x,'[]')
x = '';
end