《BP神经网络各种不同算法程序.doc》由会员分享，可在线阅读，更多相关《BP神经网络各种不同算法程序.doc（3页珍藏版）》请在金锄头文库上搜索。
1、Matlab的BP神经网络各种不同算法程序1:一般模式的BP:clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,traingd);net.trainParam.show=50;net.trainParam.lr=0.05;net.trainParam.epochs=300;net.trainParam.goal=1e-5net tr=train(net,P,T);2：加入动量的BPclcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3
2、 ,1,tansig,purelin,traingdm);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.mc=0.9;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);3：自适应LR变步长：clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,traingda);net.trainParam.show=10000
3、;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);4：弹性梯度法clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,trainrp);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net
4、.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);5：共轭梯度1clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,traincgf);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100
5、net tr=train(net,P,T);6：共轭梯度2clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,traincgp);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);7：共轭梯度3clcP=-1 -1 2 2 ;0 5 0 5;
6、T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,traincgb);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);8：共轭梯度4clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,t
7、raincgb);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);9：拟牛顿法;clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,trainbfg);net.trainParam.show=10000;net.trainParam.lr=
8、0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);10:一步正割clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,trainoss);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs
9、=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);11：levenberg-marquarat:clcP=-1 -1 2 2 ;0 5 0 5;T=-1 -1 1 1;net=newff(minmax(P),3 ,1,tansig,purelin,trainlm);net.trainParam.show=10000;net.trainParam.lr=0.05;net.trainParam.lr_inc=1.05;net.trainParam.epochs=10000;net.trainParam.goal=1e-5*100net tr=train(net,P,T);
