close all
clear
echo on
clc
pause
clc
%定義訓練樣本矢量
P=[-1:0.05:1];
randn('seed',192736547);
T=sin(2*pi*P)+0.1*randn(size(P));
pause
%繪制樣本數據點
plot(P,T,'+');
hold on;
%繪制部含噪聲的正弦曲線
plot(P,sin(2*pi*P),':');
echo on
clc
pause
%close
clc
disp('1.動量批梯度下降算法traingdm');disp('2.貝葉斯正規則算法trainbr');
choice=input('請選擇訓練算法（1，2）：');
figure(gcf);
if(choice==1)
    echo on
    clc
    %建立一個前向神經網絡    
    net=newff(minmax(P),[28,1],{'tansig','purelin'},'traingdm');
    pause
    clc
    net.trainParam.show=50;
    net.trainParam.lr=0.04;
    net.trainParam.mc=0.9;
    net.trainParam.epochs=8000;
    net.trainParam.goal=1e-3;
    net=init(net);
    pause
    clc
elseif(choice==2)
    echo on
    clc
    %建立一個前向神經網絡      
    net=newff(minmax(P),[20,1],{'tansig','purelin'},'trainbr');
    pause
    clc
    net.trainParam.show=10
    net.trainParam.epochs=5000;
    randn('seed',192736547);
    net=init(net);
    pause
    clc
end
%調用相應算法訓練BP網絡
[net,tr]=train(net,P,T);
pause
clc
%對BP網絡進行仿真
A=sim(net,P)
E=T-A;
MSE=mse(E);
pause
clc
%繪制結果曲線
close all;
plot(P,A,P,T,'+',P,sin(2*pi*P),':');
pause
clc
echo off