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！！

. e5 D" q. \+ l! E$ y%原始数据输入
clc
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,...
9 C3 P9 p% @- K% S    41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
+ j5 Z, Z  u! G. b$ vsqjdcs=[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,...
9 t4 d; G4 L4 D  F. Z  I* O    0.56,0.59,0.59,0.67,0.69,0.79];
) h7 `5 v6 x% S0 \% o9 W) D. f) Kglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
2 i; ?. {1 k' v# v* ^    22598,25107,33442,36836,40548,42927,43462];
5 P' j: y+ f1 ?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];
& L( s* N* h6 {! Nt=[glkyl;glhyl];
% d, t9 g) x. L0 J) V& L
%数据归一化
[pn,ps1]=mapminmax(p);
5 M! d. m$ q3 @[tn,ps2]=mapminmax(t);
$ ~$ T  c! z; P: ~dx=[-1,1;-1,1;-1,1];
' X+ b; v/ B0 z
%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
net.trainParam.show=1000;
. {1 Q  O! o: B: h4 [7 t+ Dnet.trainParam.Lr=0.05;
" x( E3 x, @! s- ]/ o  anet.trainParam.epochs=50000;
# Q' e/ S2 f, K! M- Dnet.trainParam.goal=0.65*10^(-3);
net=train(net,pn,tn);
- g$ L" S/ m$ n1 B
%利用原数据对BP网络仿真
; g+ [9 \) t8 q2 ?+ a8 jan=sim(net,pn);
a=mapminmax('reverse',an,ps2);
) v$ @4 x( N& m( }7 U2 U/ s* l
%仿真结果与原数据对比测试
  W" K; H5 b9 K" r- N% cx=1990:2009;
newk=a(1,:);
% N" o+ H& C/ f3 jnewh=a(2,:);
( _5 b! s' f3 x6 |8 w" zfigure(2);
% j( T% J* f+ x' O8 i  ^- lsubplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
legend('网络输出客运量','实际客运量');
6 a% e* {; J0 t. |) yxlabel('年份');ylabel('货运量、万人');
title('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--+');
! a- _9 ~1 e$ j7 S( ]legend('网络输出货运量','实际货运量');
' O" t1 q- r# d# xxlabel('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');

%新数据仿真
# }" ~2 Z, T! z  fpnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
5 z7 l" y. K, @" n3 q2 O! Wpnewn=mapminmax('apply',pnew,ps1);
anewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)
0 Q0 q& T- g! F

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
& _' h' F, x! c6 q/ o! W) Z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 ...

/ m2 T* L# S' t因为版本更新，Matlab中的Newff命令用法有所改变，原命令为：
1 H. r5 s" q8 y9 p+ Anet=newff(dx,[3,7,2]{'tansig','tansig','purelin'},'traingdx');
我修改为：
9 N8 \: R4 t- Snet=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
  a* m. j9 k9 h2 ?9 l& o& k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：
% e( ]' Y& f6 A# T[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
/ k# J5 k6 Z+ f! H4 Fa=postmnmx(an,mint,maxt);
pnewn=tramnmx(pnew,minp,maxp);
* ~. B0 ]. {/ \anew=postmnmx(anewn,mint,maxt)
我修改为：
[pn,ps1]=mapminmax(p);[tn,ps2]=mapminmax(t);
3 ^! ~0 y3 x, N0 Ca=mapminmax('reverse',an,ps2);
pnewn=mapminmax('apply',pnew,ps1);
anew=mapminmax('reverse',anewn,ps2)
2 {. ~6 b" T! M+ q, Q
4 R3 C* ~0 H, F9 S原程序为：
( {5 v! P3 b$ ~$ P%原始数据输入
) v9 Y( f8 V0 U8 ^. F, {2 mclc
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,...
' B: v1 {; o( c    41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
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,...
    2.5,2.6,2.7,2.85,2.95,3.10];
6 x5 X4 d! X3 |% K0 m. {( ?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];
3 S, K# H4 a! `. ~7 p: Y) ~glkyl=[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,...
! v' S, P4 g- F6 A6 Z9 c    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];

- r+ \5 ^9 y- a" W& I" ^%数据归一化
5 P2 ~3 m8 q0 O; O- c[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
dx=[-1,1;-1,1;-1,1];
! F. W: X# b- J7 w! D* G
9 a. E: g& K. X1 n%BP网络训练
net=newff(dx,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
5 n& I/ y4 \9 fnet.trainParam.show=1000;
net.trainParam.Lr=0.05;
net.trainParam.epochs=50000;
5 m3 J. V* b1 O5 f: }/ M& r" unet.trainParam.goal=0.65*10^(-3);
net=train(net,pn,tn);
7 B; ]5 \/ p' j# @1 e) F
%利用原数据对BP网络仿真
% a) [! z* C7 N! wan=sim(net,pn);
: k4 |3 ?# ^6 b# t* U) W  ]. P' b5 Ua=postmnmx(an,mint,maxt);

%仿真结果与原数据对比测试
x=1990:2009;
newk=a(1,;
, m; I! k! g& E" u# Vnewh=a(2,;
figure(2);
! H2 I3 W) J; S, I4 y2 asubplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
) J/ F8 C9 m0 j: O0 r; Olegend('网络输出客运量','实际客运量');
+ {4 s1 K' r5 {5 w$ M  W4 R- Txlabel('年份');ylabel('货运量、万人');
title('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--+');
% a6 A: H& K' y# V/ o: i" klegend('网络输出货运量','实际货运量');
xlabel('年份');ylabel('货运量、万吨');
$ k8 g0 g- [  O. ?3 A1 Q2 Atitle('运用工具箱货运量学习和测试对比图');

%新数据仿真
: S7 g2 b( m: l3 F0 c/ opnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
8 I3 W) I: ]9 {/ `pnewn=tramnmx(pnew,minp,maxp);
: g6 Q! v$ ?& i9 v# m* ianewn=sim(net,pnewn);
2 R! y$ D  a3 j2 Wanew=postmnmx(anewn,mint,maxt)
8 Z( V1 v5 A' A3 I; A' ?' D) c
% [  D- i) e* e" J8 ?1 h# H/ x) b修改后程序为：
%原始数据输入
clc
1 j9 [! L6 B" h" ]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];
3 f# }2 L2 i- K. [; U' E9 Ksqjdcs=[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,...
5 c: O* k- G: N6 @5 j$ |# \/ u    2.5,2.6,2.7,2.85,2.95,3.10];
, ~  @5 I/ Z5 k( D/ x2 P( [7 ssqglmj=[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];
/ h; m( p8 P) r# nglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
1 R3 S! z( S5 ^6 G    22598,25107,33442,36836,40548,42927,43462];
glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
. O, t: s4 K1 n    13320,16762,18673,20724,20803,21804];
4 ^9 l" X& q4 h) k# E! jp=[sqrs;sqjdcs;sqglmj];
8 D# L( t9 ~: k0 D. Zt=[glkyl;glhyl];

%数据归一化
[pn,ps1]=mapminmax(p);
[tn,ps2]=mapminmax(t);
; D' N( B6 v  W/ ~* i
%BP网络训练
) P" w# C, o# s8 q3 hnet=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
net.trainParam.show=1000;
net.trainParam.Lr=0.05;
net.trainParam.epochs=50000;
net.trainParam.goal=0.65*10^(-3);
net=train(net,pn,tn);

%利用原数据对BP网络仿真
an=sim(net,pn);
& \, {6 ?7 }: N2 o% n( F0 Ma=mapminmax('reverse',an,ps2);

%仿真结果与原数据对比测试
x=1990:2009;
; k* H4 ~0 _8 Z+ u6 Q4 H4 d, Unewk=a(1,;
newh=a(2,;
) I6 L$ c7 j% \1 ]+ F* i) |figure(2);
subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
+ l5 \7 c, D9 P1 t+ U9 V1 L! hlegend('网络输出客运量','实际客运量');
- y  c) G- W2 [xlabel('年份');ylabel('货运量、万人');
title('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--+');
5 w4 p9 g, d+ Z1 ~8 Slegend('网络输出货运量','实际货运量');
: y& R% s6 F1 M. Qxlabel('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');

%新数据仿真
pnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
pnewn=mapminmax('apply',pnew,ps1);
5 i+ ~; T- ~0 C% Zanewn=sim(net,pnewn);
0 F7 M  T* E" k' B8 U1 ianew=mapminmax('reverse',anewn,ps2)
6 v' x# O. e+ c: T- S: x（修改的地方用颜色标记了）
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$ S! c: \+ J4 [  ]
   
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,...
   2 F5 z4 j( q9 Z7 b) e$ o
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
   * v2 C) v$ h) Q8 v8 J) {
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,...
   4 ~( ?; ?4 l3 k) B5 O! L1 w
5.     2.5,2.6,2.7,2.85,2.95,3.10];9 j* G4 z5 @( u- A  k
   
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,...
   $ W. c, p. q) {) g
7.     0.56,0.59,0.59,0.67,0.69,0.79];
   $ [8 Q, n2 k& A9 b0 s. |7 }
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...) N2 O4 z6 d- A! m+ f9 c+ m0 T
   
9.      22598,25107,33442,36836,40548,42927,43462];4 f7 y- {% O* G& J* `' H
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    & t5 ^* |4 P\" e( v% Z* k4 F# z7 C
11.      13320,16762,18673,20724,20803,21804];
    ) f1 |2 ]% Z. f% a& s3 ]
12. p=[sqrs;sqjdcs;sqglmj];; K9 p7 z1 n+ j+ H2 C( y
    
13. t=[glkyl;glhyl];2 H  v7 Q  s; Y: Y
    
14. %数据归一化
    * B+ k( M+ L) v1 p5 \4 \
15. [pn,ps1]=mapminmax(p);0 q3 E' y, F1 x) g\" }3 N3 d
    
16. [tn,ps2]=mapminmax(t);3 ^' r5 v& B& o5 D7 l6 @
    
17. %dx=[-1,1;-1,1;-1,1];
    , O& x- [) D# Z) ]% s. i6 {
18. [color=Red] p0=minmax(pn);t0=minmax(tn);[/color]
      z- W7 x4 u7 Q3 X  G\" q* G; i; ^\" E
19. 
    8 F6 I1 J1 \2 {5 P8 \
20. %BP网络训练
    % p% g2 W5 p! F) R* c0 Y
21. [color=Red] net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); [/color]% w& Z\" L& M; T  f5 [% j1 ?
    
22. net.trainParam.show=1000;: z2 a0 l  Z. P
    
23. net.trainParam.Lr=0.05;5 i6 P3 l- @% y. ]+ K* K0 z- u
    
24. net.trainParam.epochs=50000;) l' X) i. }8 \6 Z6 a9 d- _* j
    
25. [color=Red] net.trainParam.goal=0.65*10^(-5);[/color]4 U( z0 Y6 J& Y. Y' O0 J3 U5 T
    
26. net=train(net,pn,tn);$ }7 s5 H$ d6 ?9 M
    
27. 
    $ }7 j$ f\" V* x8 Y$ T0 {
28. %利用原数据对BP网络仿真: o7 ~! `1 Z1 Q/ f( s
    
29. an=sim(net,pn);
    % P- _\" ?* N( r$ A+ K3 S
30. a=mapminmax('reverse',an,ps2);% z$ d0 w  J, Y4 d) B
    
31. 
    / h& h6 n5 j$ C4 a& D
32. %仿真结果与原数据对比测试 1 W. ~/ \0 l' _
    
33. x=1990:2009;
    & c+ p7 w4 b: t: a+ i6 d; S
34. newk=a(1,:);
    3 A9 W, w9 K' k& }6 ]  u
35. newh=a(2,:);9 x3 n9 I. U+ P8 V
    
36. figure(2);
    3 C& w! }; [; m% m1 O
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
    0 t$ y2 N) Z. f/ K
38. legend('网络输出客运量','实际客运量');
    % X6 c9 V\" b- I# g
39. xlabel('年份');ylabel('货运量、万人');( G0 u2 T9 U; Z
    
40. title('运用工具箱客运量学习和测试对比图');
    4 V  a3 s6 L7 m% L
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
    0 N- V; g  n( w6 m* `1 Z: B
42. legend('网络输出货运量','实际货运量');
    0 Y, f: ]' m\" Z$ v' o1 {. r% ?' p
43. xlabel('年份');ylabel('货运量、万吨');6 s  c* l1 A5 d5 u% }$ v4 W0 F; B% t* k
    
44. title('运用工具箱货运量学习和测试对比图');- X6 K4 O9 `* y' @: \  R, M
    
45. ; Y' ]0 |$ E2 @7 \% r, A\" s
    
46. %新数据仿真* }/ v5 ~9 d# ^
    
47. pnew=[73.39,75.552 A\" ~\" l, p; H8 C
    
48.      3.9635,4.0975
    - q$ P6 W& J) }3 A; E1 R/ D
49.      0.9880,1.0268];
    3 q0 }! h7 [, Q4 r0 C
50. pnewn=mapminmax('apply',pnew,ps1);0 z/ L- r, U, M# {. E1 C- H$ N3 E' t
    
51. anewn=sim(net,pnewn);
    % H2 I& g0 G' m& K4 i
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
   7 b\" {6 _& q4 k! \) H  Z
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,...( W/ ?\" ?: o8 F( o- [# z6 h
   
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];6 ^9 M4 `. U  j1 _9 z$ l, w
   
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,...0 i+ c9 a# g! w! o8 Q& Y
   
5.     2.5,2.6,2.7,2.85,2.95,3.10];2 C  l- _\" P3 Q' x
   
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,...
   - e; J  _8 R+ n2 E: c2 q( A
7.     0.56,0.59,0.59,0.67,0.69,0.79];
   8 {( E( {( |% U. k. J* v8 p
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...% ^( J. H8 B3 J$ M% h7 ]. H% F6 w+ o+ _
   
9.      22598,25107,33442,36836,40548,42927,43462];* w+ W' `* u$ ^4 D1 [
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
      z\" g8 f, R7 J5 n9 ?! i; q$ G
11.      13320,16762,18673,20724,20803,21804];
    - V2 Q% d6 j3 D
12. p=[sqrs;sqjdcs;sqglmj];; }8 D! I) h9 g* l9 @
    
13. t=[glkyl;glhyl];2 C1 J* a0 h) R9 {8 L) c# i& G
    
14. %数据归一化6 j/ A! i+ s' T
    
15. [pn,ps1]=mapminmax(p);  H% y1 K7 r6 @  R1 k. h
    
16. [tn,ps2]=mapminmax(t);, E: f! y4 }% W$ u; }4 |
    
17. %dx=[-1,1;-1,1;-1,1];
    - j( ?' P# E\" O; @. q' x2 p% U
18. p0=minmax(pn);t0=minmax(tn);5 N3 X$ q3 F# W+ G5 r7 M
    
19. $ Y; t3 m, o; Q# m$ W; z9 S# V/ P
    
20. %BP网络训练6 V2 e0 C4 e( k6 o# H7 N
    
21. net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); 3 |6 f+ ^$ t3 }, F; C/ e7 r
    
22. net.trainParam.show=1000;
    ! z9 _! C2 }# B3 Z! g( Y
23. net.trainParam.Lr=0.05;+ w, u; o) X0 H) o
    
24. net.trainParam.epochs=50000;/ B) L+ a; `9 @: N
    
25. net.trainParam.goal=0.65*10^(-5);
    : y' \5 B1 I- p( v
26. net=train(net,pn,tn);
    5 a$ u/ b& X8 }5 K6 ^
27. / o# i\" r( W# f! P$ w
    
28. %利用原数据对BP网络仿真9 s& X6 y- Q9 G
    
29. an=sim(net,pn);
      ?\" A! S& c5 I+ t+ R& X
30. a=mapminmax('reverse',an,ps2);
    ( ~4 a# n) ]0 ~. h) v  X
31. 9 a: S- u* M- j
    
32. %仿真结果与原数据对比测试 # F\" L7 u# l/ h) `
    
33. x=1990:2009;; u8 M. m2 K) ]% N7 a
    
34. newk=a(1,:);
    / D6 w+ [2 A* X$ n
35. newh=a(2,:);
    7 q# o, _9 j! b2 s1 L+ V9 n$ Y
36. figure(2);0 T. |( X6 N  a9 B
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');, M. |' a& m& c6 Y3 k
    
38. legend('网络输出客运量','实际客运量');2 M4 ?' p# X4 N: Q
    
39. xlabel('年份');ylabel('货运量、万人');5 e1 q* [0 z  h8 U
    
40. title('运用工具箱客运量学习和测试对比图');
    6 }4 n/ {7 D- c
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');. ~! A0 R: T( }( m  n8 G
    
42. legend('网络输出货运量','实际货运量');
    8 S$ \; o7 p+ K  u( g0 ?* O0 p
43. xlabel('年份');ylabel('货运量、万吨');4 X' F+ D- J! p$ o. p
    
44. title('运用工具箱货运量学习和测试对比图');
      S\" Z4 B/ R2 L
45. 
    - Z. o! v8 a, S. @8 E1 ^
46. %新数据仿真
    & [+ `' o# l: a4 s, E, Y
47. pnew=[73.39,75.556 m* G, r7 Z0 L9 X7 m8 a
    
48.      3.9635,4.0975\" v7 r/ m6 c# c1 O, V# h; E
    
49.      0.9880,1.0268];
    ; Q) n* S8 F6 r
50. pnewn=mapminmax('apply',pnew,ps1);
    \" P& f1 w, J! b5 u
51. anewn=sim(net,pnewn);: M, u8 h# W4 l. ^4 w) D
    
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

T-Eric

且生 发表于 2014-7-21 18:22
忽略我上面那个，改动的地方在18，21，25行……求批

: b: A+ w3 M2 o3 c$ R; }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
   # @5 d  x: H. B: `6 i
2. net=newff(p0,t0,6,{'tansig'},'traingd'); ) O, o) l: g' D( ?
   
3. net.trainParam.show=1000;5 z: L* @\" {\" U+ M. T
   
4. net.trainParam.Lr=0.05;' R( L( o! n\" s' T2 Z. p- E
   
5. net.trainParam.epochs=2000;1 a, Q0 [7 k, N) p! b+ G+ c8 i
   
6. net.trainParam.goal=0.65*10^(-4);
   - j+ t* d4 {% Z% v( d
7. net=train(net,pn,tn);

复制代码

我只改了这里面的，
, t* v5 u4 E  F8 `% R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
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 r# t9 X. H  X. y) T' U" h" H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。
5 I, m* Z5 O" E! y% L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