%--------------------------------------------------------------------------
% 利用兩層BP網(wǎng)絡(luò)完成函數(shù)逼近，其中隱層的神經(jīng)元個(gè)數(shù)選為5個(gè)
%
% initff—對(duì)前向網(wǎng)絡(luò)初始化
% simuff—對(duì)前向網(wǎng)絡(luò)仿真
% trainbp--用BP算法訓(xùn)練前向網(wǎng)絡(luò)
%--------------------------------------------------------------------------
clf;  % 從當(dāng)前圖形窗口刪除所有圖形對(duì)象，恢復(fù)圖形窗口屬性的默認(rèn)設(shè)置
figure(gcf);    % gcf--獲得當(dāng)前圖形窗口的句柄
% set(gcf,300,300);
% set(gcf, 'PaperPosition', [2 1 4 2]);
echo on;    % 允許在屏幕上顯示批處理命令行
clc;
p=-1:0.1:1;     % p為輸入矢量
t=[-0.9602  -0.577  -0.0729  0.3771  0.6405  0.66  0.4609...
        0.1336  -0.2013  -0.4344  -0.5  -0.393  -0.1647  0.0988...
        0.3072  0.396  0.3449  0.1816  -0.0312  -0.2189  -0.3201];   % t為目標(biāo)矢量
plot(p,t,'+','linewidth',2,'markersize',14);
title('Training Vectors','fontsize',24,'fontweight','bold');
xlabel('Input Vector P','fontsize',24,'fontweight','bold');
ylabel('Target Vector T','fontsize',24,'fontweight','bold');
pause;
clc;
S1=5;
[w1,b1,w2,b2]=initff(p,S1,'tansig',t,'purelin');    % 初始化神經(jīng)元，w是權(quán)值，b是閾值
echo off;
k=pickic(1);
if k==2
    w1=[3.5;  3.5;  3.5;  3.5;  3.5];
    b1=[-2.8562;  1.0774;  -0.588;  1.4083;  2.8722];
    w2=[0.2622  -0.2375  -0.4525  0.2361  -0.1718];
    b2=[0.1326];
end
echo on;
clc;
disp_freq=20;        % 學(xué)習(xí)過(guò)程顯示頻率
max_epoch=8000;     % 最大訓(xùn)練步數(shù) 
err_goal=0.02;      % 誤差指標(biāo)
lr=0.01;            % 學(xué)習(xí)率
tp=[disp_freq  max_epoch  err_goal  lr];
[w1,b1,w2,b2,epochs,tr]=trainbp(w1,b1,'tansig',w2,b2,'purelin',p,t,tp);   % 用BP算法訓(xùn)練前向網(wǎng)絡(luò)
pause;
clc;
p=0.5;    % 利用訓(xùn)練完的感知器神經(jīng)元進(jìn)行分類(lèi)
           % p=0.5時(shí)，理想輸出值=0.396
a=simuff(p,w1,b1,'tansig',w2,b2,'purelin')  % 對(duì)輸入矢量預(yù)處理
pause;
clc;
ploterr(tr,eg);    % 畫(huà)出誤差曲線(xiàn)
echo off;