BP神经网络程序问题求助

T-Eric

这是我从书上打下来的代码，因为书的运行软件版本比较低，我的是Malab2013a的版本，所以我修改了部分程序，但运行出来有问题。请哪位高手帮忙指教和修正一下。谢谢！！
  @- p/ J" o4 r
%原始数据输入
clc
3 F7 J# A& N( d& s% l; }sqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
    41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
7 X6 V4 e$ g4 K# M1 T7 U$ ?) J! ysqjdcs=[0.6,0.75,0.85,0.90,1.05,1.35,1.45,1.60,1.70,1.85,2.15,2.20,2.25,2.35,...
    2.5,2.6,2.7,2.85,2.95,3.10];
sqglmj=[0.09,0.11,0.11,0.14,0.20,0.23,0.23,0.32,0.32,0.34,0.36,0.36,0.38,0.49,...
/ F- b! n0 `! b% N; a% N8 j& q    0.56,0.59,0.59,0.67,0.69,0.79];
+ ~) H# ]* A) Z5 D$ C. M3 r1 X  fglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
    22598,25107,33442,36836,40548,42927,43462];
glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
" J" l6 Y  g0 `8 g4 E+ H    13320,16762,18673,20724,20803,21804];
5 B  r6 c3 v# [. Q0 W2 |" Gp=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];
3 n) s  Q5 @: g7 u
; g- S# B7 s( s) O" Z6 Y# h%数据归一化
[pn,ps1]=mapminmax(p);
[tn,ps2]=mapminmax(t);
dx=[-1,1;-1,1;-1,1];
4 H. _, R9 ^" Q* [
% @% v; B, L% I%BP网络训练
* b/ i2 d: r( U* ]net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
6 T3 u/ o; y2 l7 |* J& {net.trainParam.show=1000;
net.trainParam.Lr=0.05;
! J" A2 b1 x/ J  W8 I& |( x6 vnet.trainParam.epochs=50000;
net.trainParam.goal=0.65*10^(-3);
+ J* d( T' y4 ?4 F3 ynet=train(net,pn,tn);
2 X+ _! N  I( Z9 C7 A
%利用原数据对BP网络仿真
9 K. W8 S) u; e, x6 }$ u* }9 nan=sim(net,pn);
a=mapminmax('reverse',an,ps2);

%仿真结果与原数据对比测试
x=1990:2009;
newk=a(1,:);
newh=a(2,:);
figure(2);
subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
legend('网络输出客运量','实际客运量');
xlabel('年份');ylabel('货运量、万人');
* o0 c' Z# Z/ j. Etitle('运用工具箱客运量学习和测试对比图');
- R! b6 a# m0 P" u1 Q# lsubplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
legend('网络输出货运量','实际货运量');
8 h% v; ~4 I7 Y3 X8 _xlabel('年份');ylabel('货运量、万吨');
2 C, @4 v: K( J# i! ?( h$ ktitle('运用工具箱货运量学习和测试对比图');

# I7 w7 p0 q2 A  p0 K3 m5 I%新数据仿真
3 P  ?5 c) V8 q/ V7 o- bpnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
' {+ D9 i' t9 @8 Lpnewn=mapminmax('apply',pnew,ps1);
anewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)

% C  ~7 m% O5 b8 Z# ~# Q5 f! O( B) k

gancm

看了图我感觉是网络输出后面那几个值太大了，前面基本都是一条水平线了，具体哪里还看不出问题。你把修改的地方说一下看看

529084167

T-Eric

gancm 发表于 2014-7-19 12:30
- g" p$ p" x$ H' p6 o看了图我感觉是网络输出后面那几个值太大了，前面基本都是一条水平线了，具体哪里还看不出问题。你把修改的 ...

" @' p& X- E1 k$ @9 E因为版本更新，Matlab中的Newff命令用法有所改变，原命令为：
net=newff(dx,[3,7,2]{'tansig','tansig','purelin'},'traingdx');
; t  W& o3 A4 S2 B$ r我修改为：
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
1 J( M; F' e3 X5 v然后有些命令被其他命令替代了，其中有premnmx，postmnmx，tramnmx命令好像被替换了，使用了一个强大的命令mapminmax。原命令为：
[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
a=postmnmx(an,mint,maxt);
pnewn=tramnmx(pnew,minp,maxp);
anew=postmnmx(anewn,mint,maxt)
  s" b( O$ U4 N- D0 e我修改为：
[pn,ps1]=mapminmax(p);[tn,ps2]=mapminmax(t);
a=mapminmax('reverse',an,ps2);
pnewn=mapminmax('apply',pnew,ps1);
& j" `# k( g$ j' _" D" \2 Q+ yanew=mapminmax('reverse',anewn,ps2)
- z0 A" l/ z: ]2 N0 K: @# x2 p
原程序为：
%原始数据输入
, U& \3 z8 `: }, _1 ]+ wclc
) _, @4 U) n4 P0 B& Q+ X3 ssqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
    41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
2 b4 S' ~3 N& F* v( nsqjdcs=[0.6,0.75,0.85,0.90,1.05,1.35,1.45,1.60,1.70,1.85,2.15,2.20,2.25,2.35,...
/ X( v1 l1 J3 S3 g    2.5,2.6,2.7,2.85,2.95,3.10];
sqglmj=[0.09,0.11,0.11,0.14,0.20,0.23,0.23,0.32,0.32,0.34,0.36,0.36,0.38,0.49,...
    0.56,0.59,0.59,0.67,0.69,0.79];
glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
- U& b- n% r4 m# j4 g$ K% k    22598,25107,33442,36836,40548,42927,43462];
glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];
& w% m- s. t2 q: _
%数据归一化
[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
& g6 ?; r7 n& Jdx=[-1,1;-1,1;-1,1];
1 a# b- k5 S! w7 Q3 L
%BP网络训练
net=newff(dx,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
4 U. ?8 y6 Z  ]0 v  lnet.trainParam.show=1000;
net.trainParam.Lr=0.05;
( V: q' ^9 m, w( {) U* S) o! pnet.trainParam.epochs=50000;
net.trainParam.goal=0.65*10^(-3);
0 X' S' d9 G/ [! hnet=train(net,pn,tn);

%利用原数据对BP网络仿真
an=sim(net,pn);
a=postmnmx(an,mint,maxt);

4 ~5 M5 O& v; I8 n, [%仿真结果与原数据对比测试
: k; \# X) P( z# Mx=1990:2009;
. m  M9 b* d" D2 q1 i8 K7 ?; |7 q1 Bnewk=a(1,;
newh=a(2,;
7 p/ Y& b9 E5 p4 qfigure(2);
subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
legend('网络输出客运量','实际客运量');
xlabel('年份');ylabel('货运量、万人');
- q+ W, R. Y: E! Rtitle('运用工具箱客运量学习和测试对比图');
& d2 [* g/ E! c$ ]8 Zsubplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
" R+ L( z9 F" O; f9 ulegend('网络输出货运量','实际货运量');
2 ^& Z$ P; j- ?+ V% j& Y7 Yxlabel('年份');ylabel('货运量、万吨');
title('运用工具箱货运量学习和测试对比图');
% m; n! I# U2 q
( _- B' r' r' l* t' A. Z%新数据仿真
pnew=[73.39,75.55
+ j0 _) {- i0 M1 O    3.9635,4.0975
    0.9880,1.0268];
pnewn=tramnmx(pnew,minp,maxp);
anewn=sim(net,pnewn);
anew=postmnmx(anewn,mint,maxt)

& p6 a$ n0 o% p修改后程序为：
/ `9 y' A, Y2 }" d) S/ |%原始数据输入
clc
  T" c4 o" [# R0 p, P( P1 f4 rsqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
    41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
- h; B* q4 {; k! Bsqjdcs=[0.6,0.75,0.85,0.90,1.05,1.35,1.45,1.60,1.70,1.85,2.15,2.20,2.25,2.35,...
    2.5,2.6,2.7,2.85,2.95,3.10];
sqglmj=[0.09,0.11,0.11,0.14,0.20,0.23,0.23,0.32,0.32,0.34,0.36,0.36,0.38,0.49,...
# |# `; e0 n6 y    0.56,0.59,0.59,0.67,0.69,0.79];
  @- h5 h0 p& K2 Aglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
    22598,25107,33442,36836,40548,42927,43462];
glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
  n) L9 k$ ~- Ot=[glkyl;glhyl];

%数据归一化
[pn,ps1]=mapminmax(p);
5 o& F0 J' |. g. M9 e[tn,ps2]=mapminmax(t);
! v* Q% k3 C/ f) O2 d5 C" Q/ C# g
+ Z1 O( P9 H! H$ P- d$ z, j%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
3 o/ p3 k9 T8 q$ vnet.trainParam.show=1000;
& a3 R& @; u( t9 Onet.trainParam.Lr=0.05;
# ^" U) X. q2 I  y8 @$ U/ l* xnet.trainParam.epochs=50000;
net.trainParam.goal=0.65*10^(-3);
$ G2 E3 {% c# Tnet=train(net,pn,tn);

( j! ?: |/ O- S7 p1 X%利用原数据对BP网络仿真
an=sim(net,pn);
% ~# H9 _9 L! L; l- d' W! D6 Va=mapminmax('reverse',an,ps2);

%仿真结果与原数据对比测试
x=1990:2009;
7 A% I6 t& S) o" O# Q& jnewk=a(1,;
1 r5 b$ i$ C+ }  u- `' s$ n3 N. _newh=a(2,;
* ]+ ^0 Q: o; |+ u% afigure(2);
, u4 j, S6 ?. k' l% t- _. F# rsubplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
legend('网络输出客运量','实际客运量');
" |  a7 D# t7 l* @( A" ]/ pxlabel('年份');ylabel('货运量、万人');
; h3 d6 J) I! y; ctitle('运用工具箱客运量学习和测试对比图');
subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
$ C7 P8 A) c4 f$ L, Wlegend('网络输出货运量','实际货运量');
xlabel('年份');ylabel('货运量、万吨');
# X) h1 ]# l9 i4 K# }# utitle('运用工具箱货运量学习和测试对比图');
' D5 w6 f$ A1 D( X/ Y" }/ k2 X
%新数据仿真
pnew=[73.39,75.55
- i8 A0 h0 R. U" F% W& c    3.9635,4.0975
    0.9880,1.0268];
pnewn=mapminmax('apply',pnew,ps1);
anewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)
, c) e  l" ?0 }- I（修改的地方用颜色标记了）
$ F% P7 C- q0 W$ [+ V$ w# p麻烦您帮忙指出其中的问题，万分感谢！

且生

求书名及页码……

且生

1. clc# S2 W/ {' c9 Q% [5 h\" s: S* t
   
2. sqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...4 B( g- H* w; e$ {, M! \% Z0 f
   
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];- k  z) P- f3 w4 }+ A& o, F4 C0 J4 L
   
4. sqjdcs=[0.6,0.75,0.85,0.90,1.05,1.35,1.45,1.60,1.70,1.85,2.15,2.20,2.25,2.35,...% K: t2 j; ]7 N3 ?
   
5.     2.5,2.6,2.7,2.85,2.95,3.10];
   / W$ j0 c$ h* o6 z\" ^
6. sqglmj=[0.09,0.11,0.11,0.14,0.20,0.23,0.23,0.32,0.32,0.34,0.36,0.36,0.38,0.49,...
   * {; g# R% X( I9 j* \! ^0 N
7.     0.56,0.59,0.59,0.67,0.69,0.79];2 }' C' x& o0 j$ J( ]/ A, P
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
   8 A4 }$ _# c* u; q& T1 T
9.      22598,25107,33442,36836,40548,42927,43462];
   # j& n) B/ ~* _! }' |- w
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    \" [6 P4 m- b0 J  c5 F; \1 N
11.      13320,16762,18673,20724,20803,21804];9 M& Q$ {* B  a( Q) i
    
12. p=[sqrs;sqjdcs;sqglmj];
    , s, i% x6 X5 @+ l3 \: E1 h# E) d9 G) l% v
13. t=[glkyl;glhyl];6 T5 o3 F1 e4 `3 t$ o, a
    
14. %数据归一化4 f6 z8 }$ n) Z: ]\" j/ `9 M
    
15. [pn,ps1]=mapminmax(p);
    # X7 g; e. N- j' T. W, E\" S
16. [tn,ps2]=mapminmax(t);
    ! F& L6 }\" O* U0 k3 B. n
17. %dx=[-1,1;-1,1;-1,1];
    ( @# p% i  O\" ]
18. [color=Red] p0=minmax(pn);t0=minmax(tn);[/color]
    2 Q. {$ Y& m9 w; p9 o
19. 
    6 H  M( x/ t2 {0 i  [  M6 P
20. %BP网络训练5 u5 J8 @) G# C& p, T& W9 I' _
    
21. [color=Red] net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); [/color]
    * U1 \$ [) G( y% B: L( E
22. net.trainParam.show=1000;
    $ [0 u% U3 {- h7 o, t
23. net.trainParam.Lr=0.05;
      P4 F, u! ]. e) w- Q
24. net.trainParam.epochs=50000;' c9 e' I# A) I/ C! x
    
25. [color=Red] net.trainParam.goal=0.65*10^(-5);[/color]$ }7 p& b2 L. W7 D
    
26. net=train(net,pn,tn);
    1 S\" V: ?6 @( W+ B& H
27. 
    : l\" I0 z) s! }4 E
28. %利用原数据对BP网络仿真4 ~6 t; n8 T7 q. Z9 O7 f3 F
    
29. an=sim(net,pn);& I( x- k( E* ?1 z! x
    
30. a=mapminmax('reverse',an,ps2);$ E' y) P. k0 t% ~3 D; V8 J9 X/ N
    
31. * ^% z\" }; x0 K' ?) B7 B5 u0 p, P
    
32. %仿真结果与原数据对比测试 0 b5 q  Z2 E0 [
    
33. x=1990:2009;; N+ z/ ^) `$ q# v7 F1 N! }3 V
    
34. newk=a(1,:);
    $ h3 W8 S/ M- Q4 |
35. newh=a(2,:);
    ! i1 @' ]7 H  o. Q+ q! A9 F; `
36. figure(2);2 K' D\" V/ o5 J& ]# [1 T/ t3 m
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');! Q+ T- W% Y2 A; p/ J
    
38. legend('网络输出客运量','实际客运量');1 u/ {! G; N7 c; k' b% A
    
39. xlabel('年份');ylabel('货运量、万人');, |/ c* Y5 q4 J, Y/ x8 }6 k# G
    
40. title('运用工具箱客运量学习和测试对比图');
    6 y& V1 a6 W, k% P0 n) J5 X
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
    ! M% n' M\" G\" D* z1 p1 [7 |
42. legend('网络输出货运量','实际货运量');
    : b5 e5 l4 \5 A1 x
43. xlabel('年份');ylabel('货运量、万吨');, r\" ?5 w5 ^1 O+ q# [! g- s
    
44. title('运用工具箱货运量学习和测试对比图');1 s# c4 |' w! A
    
45. 
    6 D7 \# L8 N# V# U6 X
46. %新数据仿真- d+ @# i( w# O, ~' I( l# ~# O
    
47. pnew=[73.39,75.55
    7 C# ^* F! G$ G: h\" |
48.      3.9635,4.0975
    & A3 V! B. C) s$ b5 `
49.      0.9880,1.0268];
    / E* a* L* E+ ?5 U: d6 t  m
50. pnewn=mapminmax('apply',pnew,ps1);. e1 M7 z) `\" k) h
    
51. anewn=sim(net,pnewn);
    . N8 q  j8 r& o5 n' V
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

且生

忽略我上面那个，改动的地方在18，21，25行……求批

1. clc
   2 B# g* |9 C5 R1 R$ K
2. sqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
   ' f% G2 |- e% D& n3 h
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
   $ t  Z2 T0 z\" j  c& ]5 m: t
4. sqjdcs=[0.6,0.75,0.85,0.90,1.05,1.35,1.45,1.60,1.70,1.85,2.15,2.20,2.25,2.35,..., a+ q0 y/ R: c, K( \. u' u  G8 n
   
5.     2.5,2.6,2.7,2.85,2.95,3.10];
   . z& p# Y  z$ _8 S/ Z0 O
6. sqglmj=[0.09,0.11,0.11,0.14,0.20,0.23,0.23,0.32,0.32,0.34,0.36,0.36,0.38,0.49,...1 u1 E, `- D$ F9 o/ l\" g
   
7.     0.56,0.59,0.59,0.67,0.69,0.79];. R/ m0 p  ~- B* b% Z+ k2 a- f% L
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
   % v2 ^* P+ u6 X1 ]
9.      22598,25107,33442,36836,40548,42927,43462];8 f& c+ {: X3 c5 c\" F
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    4 l! N% l2 ?, `* Y  @+ m
11.      13320,16762,18673,20724,20803,21804];( ?$ d  t- f, T- v9 @- C
    
12. p=[sqrs;sqjdcs;sqglmj];: y0 C0 G\" R5 I5 F\" @% Y9 X7 o
    
13. t=[glkyl;glhyl];/ w: L5 A\" ]7 V2 T) L
    
14. %数据归一化' b/ p3 m5 X1 L( l! T. H
    
15. [pn,ps1]=mapminmax(p);$ R) {- h# _& z: ]. G1 O
    
16. [tn,ps2]=mapminmax(t);
    & g4 }4 L3 U0 v
17. %dx=[-1,1;-1,1;-1,1];* |+ B0 Q  ]  I% ]$ N  E5 l
    
18. p0=minmax(pn);t0=minmax(tn);
      P3 a$ m6 r\" {, [* X0 S
19. 
    8 X3 [3 B( C) U/ |/ i1 o
20. %BP网络训练
    ; r0 Q4 o$ p* x* o( V- l& {
21. net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
    / ^7 }9 j. h) z& a  U, R. C
22. net.trainParam.show=1000;
    : e1 A, z$ O0 @- q\" @
23. net.trainParam.Lr=0.05;
    6 L- I5 W' k& P2 ]
24. net.trainParam.epochs=50000;
    + O. g) |) X$ c# @0 J\" D
25. net.trainParam.goal=0.65*10^(-5);5 r* Y* {7 R! X/ N
    
26. net=train(net,pn,tn);
    ' `3 l' N. V& Z2 G3 n
27. 0 E' j5 v+ ^( J- g9 g
    
28. %利用原数据对BP网络仿真
    ' Y8 P2 Q$ ^: t& t' I4 K
29. an=sim(net,pn);
    . u\" U( h- f1 }  T; a
30. a=mapminmax('reverse',an,ps2);
    ! L# F7 z. _2 x) @) N  c* R2 l/ C
31. 8 t( V! Q- R  k8 k. i
    
32. %仿真结果与原数据对比测试
    * }1 E6 G9 E/ C6 f  }( r8 t
33. x=1990:2009;
    \" T/ H3 `$ H8 }8 X5 ?
34. newk=a(1,:);
    . {6 c  ]2 }\" s& e0 l& k9 X
35. newh=a(2,:);. ^. \! D! @8 E8 y& V! K0 l
    
36. figure(2);# `* L* S* p5 S2 u- d& r
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
    $ H+ D% |0 P7 Z8 B' P  [\" h
38. legend('网络输出客运量','实际客运量');+ S7 T5 H7 v  [6 I. T( t' D\" G
    
39. xlabel('年份');ylabel('货运量、万人');
    ( \, o/ n% X2 L\" ]
40. title('运用工具箱客运量学习和测试对比图');
    ( g$ N& O$ J  D. z
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
    3 O4 K\" B5 r* w; o
42. legend('网络输出货运量','实际货运量');! \2 t4 M: d% u2 [& s( X
    
43. xlabel('年份');ylabel('货运量、万吨');3 q/ D4 @2 y+ t) N+ ]
    
44. title('运用工具箱货运量学习和测试对比图');( a& G+ I8 h5 l7 K% D: K7 V+ W
    
45. / |( c* h9 y3 h
    
46. %新数据仿真
    ( ^* t4 S8 y* c3 c
47. pnew=[73.39,75.554 u8 c6 T2 h8 _- M
    
48.      3.9635,4.0975
    : A/ n5 J\" f: J% }
49.      0.9880,1.0268];; t5 ?\" Z( J8 I1 O. h& J$ ^' f4 w
    
50. pnewn=mapminmax('apply',pnew,ps1);
    , y% v3 O5 N1 G; o+ s+ v6 `) m
51. anewn=sim(net,pnewn);
    % Y$ E: W, I, }2 y! N1 i2 k; r
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

T-Eric

且生 发表于 2014-7-21 18:22
4 t( E, E! {5 |& N6 R忽略我上面那个，改动的地方在18，21，25行……求批

5 N2 `9 j7 a2 h  j' ]0 [确实改善了很多，很是感谢。但效果还是不太理想，而且为何只学习了几十次就停了呢？即使我修改了目标精度。

且生

1. %BP网络训练
   ! g! z) B' B+ t( ^
2. net=newff(p0,t0,6,{'tansig'},'traingd'); $ s\" d1 m. L( P* n7 M) ]
   
3. net.trainParam.show=1000;
   ! K\" |3 n' n3 p2 S4 T+ F8 s. L
4. net.trainParam.Lr=0.05;' q4 t4 ]+ N/ d- j1 [  S( F
   
5. net.trainParam.epochs=2000;% W! D\" X8 ?3 z+ X3 k
   
6. net.trainParam.goal=0.65*10^(-4);
   ( z% _\" l- a& u' x7 ?: j# }
7. net=train(net,pn,tn);

复制代码

我只改了这里面的，
8 T  X& H5 V: l, e5 \一、改成单隐含层的，6个节点
二、训练函数改成梯度下降BP算法 traingd
! G: l. e1 B* o; ^9 L, A三、迭代次数改成2000
上面的参数是自己试的，我也不知道为什么。
8 m, o( N  p, ]+ G; k; l* c由于这玩意儿比较不靠谱，楼主多运行几次就能找到拟合的比较好的网络。
关键问题是有没有过拟合我也不知道，等大神来解答吧

且生

因为样本比较少，所以训练函数没有选会调节学习率的