神经网络预测.zip_combinationynm_神经网络智能预测_神经网预测

function varargout = NN(varargin) % NN MATLAB code for NN.fig % NN, by itself, creates a new NN or raises the existing % singleton*. % % H = NN returns the handle to a new NN or the handle to % the existing singleton*. % % NN('CALLBACK',hObject,eventData,handles,...) calls the local % function named CALLBACK in NN.M with the given input arguments. % % NN('Property','Value',...) creates a new NN or raises the % existing singleton*. Starting from the left, property value pairs are % applied to the GUI before NN_OpeningFcn gets called. An % unrecognized property name or invalid value makes property application % stop. All inputs are passed to NN_OpeningFcn via varargin. % % *See GUI Options on GUIDE's Tools menu. Choose "GUI allows only one % instance to run (singleton)". % % See also: GUIDE, GUIDATA, GUIHANDLES % Edit the above text to modify the response to help NN % Last Modified by GUIDE v2.5 15-Dec-2017 21:31:38 % Begin initialization code - DO NOT EDIT gui_Singleton = 1; gui_State = struct('gui_Name', mfilename, ... 'gui_Singleton', gui_Singleton, ... 'gui_OpeningFcn', @NN_OpeningFcn, ... 'gui_OutputFcn', @NN_OutputFcn, ... 'gui_LayoutFcn', [] , ... 'gui_Callback', []); if nargin && ischar(varargin{1}) gui_State.gui_Callback = str2func(varargin{1}); end if nargout [varargout{1:nargout}] = gui_mainfcn(gui_State, varargin{:}); else gui_mainfcn(gui_State, varargin{:}); end % End initialization code - DO NOT EDIT % --- Executes just before NN is made visible. function NN_OpeningFcn(hObject, eventdata, handles, varargin) % This function has no output args, see OutputFcn. % hObject handle to figure % eventdata reserved - to be defined in a future version of MATLAB % handles structure with handles and user data (see GUIDATA) % varargin command line arguments to NN (see VARARGIN) % Choose default command line output for NN handles.output = hObject; % Update handles structure guidata(hObject, handles); % UIWAIT makes NN wait for user response (see UIRESUME) % uiwait(handles.figure1); % --- Outputs from this function are returned to the command line. function varargout = NN_OutputFcn(hObject, eventdata, handles) % varargout cell array for returning output args (see VARARGOUT); % hObject handle to figure % eventdata reserved - to be defined in a future version of MATLAB % handles structure with handles and user data (see GUIDATA) % Get default command line output from handles structure varargout{1} = handles.output; % --- Executes on button press in pushbutton1. function pushbutton1_Callback(hObject, eventdata, handles) % hObject handle to pushbutton1 (see GCBO) % eventdata reserved - to be defined in a future version of MATLAB % handles structure with handles and user data (see GUIDATA) NumOfSample=str2num(get(handles.edit1,'string'));%样本个数 g=str2num(get(handles.edit2,'string'));%训练步长 z=str2num(get(handles.edit3,'string'));%阻尼因子 NE=str2num(get(handles.edit4,'string'));%训练次数 l=str2num(get(handles.edit5,'string'));%遗忘因子1 a=str2num(get(handles.edit6,'string'));%遗忘因子2 A={'radiobutton1','radiobutton2','radiobutton3','radiobutton4',}; B={'BP算法','改进BP算法','Davidon最小二乘法','阻尼最小二乘法'}'; str1=handles.select1; C={'radiobutton5','radiobutton6','radiobutton7',}; D={'随机数','正弦信号','方波信号'}'; str2=handles.select2; for j=1:1:3 if(strcmp(str2,C{j})) if j==1 u=rand(1,101);%随机数 end if j==2 t=1:101; u=sin(t); %正弦 end if j==3 u=5*[zeros(100/4,1);-ones(100/4,1);zeros(100/4,1);-ones(100/4+1,1)];%方波 end end end for j=1:1:4 if(strcmp(str1,A{j})) if j==1 % NumOfSample=100; %NumOfHP为隐藏节点数 NumOfHP=8; %g为训练速率 %g=-0.2; %NumOfEpoch为训练次数 %NE=1000; NumOfEpoch=NumOfSample*NE; %原系统的输入量u(个数为NumOfSample) %for m=1:NumOfSample+1 % u(m)=sin(m); %end %u=rand(1,101); %u=rand(1,NumOfSample+1); %原系统的理想输出量y y(1)=0; y(2)=0; for m=3:NumOfSample+2 y(m)=(0.8-0.5*exp(-(y(m-1))^2))*y(m-1)-(0.3+0.9*exp(-(y(m-1))^2))*y(m-2)... +u(m-1)+0.2*u(m-2)+0.1*u(m-1)*u(m-2); end %下面建立初始网络，4个输入节点，NumOfHP个隐藏节点，1个输出节点 %设定输入层与隐藏层之间的默认初始权值矩阵 V_ih，阈值为1 V_ih=2*(2*rand(NumOfHP,5)-1); %设定隐藏层与输出层之间的默认初始权值矩阵W_ji，阈值为1 W_ji=2*(2*rand(1,NumOfHP+1)-1); %下面训练开始神经网络 k=1; while k<=NumOfEpoch k_=k; k_=mod(k_,NumOfSample); if k_==0 k_=NumOfSample; end %给出输入节点S_h的值 S_h=[y(k_),y(k_+1),u(k_),u(k_+1),1]; %给出输出目标D_j D_j=y(k_+2); %求隐藏层的输出 Net_Ai=S_h* V_ih'; for i=1:NumOfHP A_i(i)=1/(1+exp(-Net_Ai(i))); end A_i(NumOfHP+1)=1; %求输出节点的输出 R_j=A_i*W_ji'; %求误差 e=D_j-R_j; %调整输入层与隐藏层之间的连接权值和阈值 for i=1:NumOfHP for h=1:5 V_ih(i,h)=V_ih(i,h)-g*e*W_ji(1,i)*(1-(A_i(i)))*A_i(i)*S_h(h); end end %调整隐藏层与输出层之间的连接权值和阈值 for i=1:NumOfHP+1 W_ji(1,i)=W_ji(1,i)-g*e*A_i(i); end k=k+1; end %下面为仿真验证 uu=u; %uu=rand(1,101); yy(1)=0; yy(2)=0; for m=3:102 yy(m)=(0.8-0.5*exp(-(yy(m-1))^2))*yy(m-1)-(0.3+0.9*exp(-(yy(m-1))^2))*yy(m-2)... +uu(m-1)+0.2*uu(m-2)+0.1*uu(m-1)*uu(m-2); end for kk=1:100 SS_h=[yy(kk),yy(kk+1),uu(kk),uu(kk+1),1]; DD_j(kk)=yy(kk+2); NNet_Ai=SS_h* V_ih'; for i=1:NumOfHP AA_i(i)=1/(1+exp(-NNet_Ai(i))); end AA_i(NumOfHP+1)=1; RR_j(kk)=AA_i*W_ji'; end for t=1:100 a(t)=yy(t+2); b(t)=RR_j(t); c(t)=a(t)-b(t); end for t=1:100; set(handles.edit7,'string',t)%动态显示运行步数 t=1:t axes(handles.axes1); plot(t,a(t),'r') hold on plot(t,b(t),'b') legend('y','y^'); hold off %figure(2); axes(handles.axes2); plot(t,c(t),'g') pause(0.1);%动态绘图 hold on end end if j==2 % NumOfSample=100; %NumOfHP为隐藏节点个数 NumOfHP=8; %g为训练速率 %g=-0.2; %a为遗忘因子 %a=0.2; %NumOfEpoch为训练次数 %NE=100; NumOfEpoch=NumOfSample*NE; %原系统的输入量u(个数为NumOfSample) %for m=1:NumOfSample+1 %u(m)=sin(m); %end %u=rand(1,NumOfSample+1); %原系统的理想输出量y y(1)=0; y(2)=0; for m=3:NumOfSample+2 y(m)=(0.8-0.5*exp(-(y(m-1))^2))*y(m-1)-(0.3+0.9*exp(-(y(m-1))^2))*y(m-2)... +u(m-1)+0.2*u(m-2)+0.1*u(m-1