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！！

& c0 i5 C4 j6 X9 E0 d2 o%原始数据输入
: R. N* A  R( J- \( H/ K& I3 vclc
7 r: ]7 F. p. osqrs=[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];
- |* W, M5 }7 P, p  {  D+ fsqjdcs=[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 _4 I" f* |5 r4 H" ~; @8 M    2.5,2.6,2.7,2.85,2.95,3.10];
7 \$ f  s5 B3 S- o( M8 B( Hsqglmj=[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,...
3 T( v* C  [4 ~) \    0.56,0.59,0.59,0.67,0.69,0.79];
% ~5 C! d) |/ T3 Vglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
    22598,25107,33442,36836,40548,42927,43462];
, |0 w9 e' O: u' Rglhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
# G2 R2 D3 c2 _, Q5 w& R2 s    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];

%数据归一化
& h( i! H3 S6 N[pn,ps1]=mapminmax(p);
8 M0 A! z7 t, C1 b7 @$ x) L[tn,ps2]=mapminmax(t);
dx=[-1,1;-1,1;-1,1];
8 s1 N; s: ^, N# j* a: S
%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
4 b% D) H  ~$ R' j1 H9 ?  S, K, Gnet.trainParam.show=1000;
2 \8 k5 d3 M* J8 o. {+ bnet.trainParam.Lr=0.05;
  F$ P4 }3 y* E" wnet.trainParam.epochs=50000;
net.trainParam.goal=0.65*10^(-3);
net=train(net,pn,tn);
) K- w% A  r& U
- Z* r$ B, \6 q  ?, k  e%利用原数据对BP网络仿真
0 B8 z4 ]2 C: o! Gan=sim(net,pn);
6 a& s: i" x  d% W& r& ~5 h* ~a=mapminmax('reverse',an,ps2);
" k+ d7 T2 l; [7 Y1 I$ _9 R! Z8 m
%仿真结果与原数据对比测试
- c4 e8 m3 n" f# I) |. p: Fx=1990:2009;
newk=a(1,:);
newh=a(2,:);
% ]5 b! [( d$ X% d5 c" Pfigure(2);
& L- I& T* x  c( n9 V2 s3 Ksubplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
5 U) v1 k" c$ B& n- Ylegend('网络输出客运量','实际客运量');
xlabel('年份');ylabel('货运量、万人');
6 g& Y1 F& O  d9 s% G; u' U+ n& Wtitle('运用工具箱客运量学习和测试对比图');
2 t0 h5 x# K7 G0 P3 y" ]subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
legend('网络输出货运量','实际货运量');
xlabel('年份');ylabel('货运量、万吨');
- `5 @" d3 U8 Z# Atitle('运用工具箱货运量学习和测试对比图');

%新数据仿真
$ {" P  s: {9 T' B& I+ _pnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
! }0 N! @2 N* {8 e% _+ H, bpnewn=mapminmax('apply',pnew,ps1);
; Z  _# B  x' n8 u' U( D5 Manewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)
3 \- b. [1 s4 C( V$ Z
1 L3 }9 ?/ m4 l, j$ J( q) J

T-Eric

狼之魂汪洋 发表于 2014-8-6 14:00
2 L. p' v  Z  v你这例题是在什么书上找的？

就是那本《matalb在数学建模中的应用》

狼之魂汪洋

你这例题是在什么书上找的？

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

且生

1. %BP网络训练/ Y6 K& G) {! U
   
2. net=newff(p0,t0,6,{'tansig'},'traingd'); 0 k8 k) ?. z9 T; S: Y
   
3. net.trainParam.show=1000;$ @% p* l5 {7 b
   
4. net.trainParam.Lr=0.05;! k+ I6 o. Y0 S' {
   
5. net.trainParam.epochs=2000;7 x1 V& g& u1 e
   
6. net.trainParam.goal=0.65*10^(-4);
   % n# e8 _1 |- \$ Y/ i
7. net=train(net,pn,tn);

复制代码

我只改了这里面的，
, q7 c5 e. {8 q. m一、改成单隐含层的，6个节点
7 J# Y1 E; N9 z' [  v$ z0 H9 j2 ~二、训练函数改成梯度下降BP算法 traingd
8 x( N- j/ p6 U5 t5 q/ V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。
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

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

8 f% t& F7 H  o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

忽略我上面那个，改动的地方在18，21，25行……求批

1. clc) Z1 O\" i9 H! D
   
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,...
   3 L$ w4 @$ R, m1 G
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];  ~! h  ^0 _\" j/ ^0 G: o1 X1 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,...+ H\" T- m9 ^' Q\" D8 X6 g( ^( H
   
5.     2.5,2.6,2.7,2.85,2.95,3.10];
   / p9 S, W% }* T
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,...9 {( V* M9 w$ i
   
7.     0.56,0.59,0.59,0.67,0.69,0.79];) V# n; `# k$ m/ \. @4 @. o4 K
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...) \! b& H  ?* _+ V
   
9.      22598,25107,33442,36836,40548,42927,43462];& U5 R# t\" x5 ?, R6 N5 N* a
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...5 B2 W) r1 v# P' G* M! T- X
    
11.      13320,16762,18673,20724,20803,21804];
    + E2 C) r0 s) V. R5 J
12. p=[sqrs;sqjdcs;sqglmj];
    6 _\" q0 U  t7 i: J) C% ^/ }/ {& u. i
13. t=[glkyl;glhyl];4 ?  |1 h\" m% j/ [$ |5 y9 D# U: _4 l
    
14. %数据归一化8 q3 ]3 y, U  I, D' l; F\" N
    
15. [pn,ps1]=mapminmax(p);
      s. m+ q& C, T9 x$ x- _; l
16. [tn,ps2]=mapminmax(t);+ c7 d4 o7 c3 U9 f* A9 S
    
17. %dx=[-1,1;-1,1;-1,1];
    4 k5 L\" k) y. A* t5 m2 D2 X
18. p0=minmax(pn);t0=minmax(tn);4 q$ U$ V# \2 |3 E
    
19. 
    / |. N) P( U' G0 _\" r6 s& ~' w& G2 b
20. %BP网络训练
    ! S\" C. g* u' i5 [6 u$ o- `
21. net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); 1 S' f\" y* P2 a& p6 s
    
22. net.trainParam.show=1000;
    - [3 T$ x6 z$ G% \+ r
23. net.trainParam.Lr=0.05;
    6 `) S( B8 z/ k5 p$ t7 ?
24. net.trainParam.epochs=50000;
    2 Z$ o8 x$ }# C2 H5 e
25. net.trainParam.goal=0.65*10^(-5);
    0 g* `# o3 N5 G: Y
26. net=train(net,pn,tn);\" W) `; [* |\" D' e' B# E
    
27. 
      u9 a1 D# @+ {. e) ?0 e. x1 X
28. %利用原数据对BP网络仿真
    % o; M0 ?\" [  Y  K
29. an=sim(net,pn);
    ; W3 c0 N) b$ B& v
30. a=mapminmax('reverse',an,ps2);
    / K& }+ w- R' f- R
31. 0 ~# H; I0 ]8 b8 E! b5 w  f7 v
    
32. %仿真结果与原数据对比测试 : T5 j* m3 Y& j! \$ n' q
    
33. x=1990:2009;% W) ]) Q1 U6 {: ?
    
34. newk=a(1,:);7 E; x! B- F5 L, L
    
35. newh=a(2,:);1 K- ?( M# J* ~1 A
    
36. figure(2);
    ! x& {% c2 L# ]4 s3 `+ l
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');  s' H( T  m2 `$ _! o. U; h
    
38. legend('网络输出客运量','实际客运量');2 {* v% c7 n* y/ u8 _
    
39. xlabel('年份');ylabel('货运量、万人');
    ) W! J7 J/ ]5 E; V& \. {4 E
40. title('运用工具箱客运量学习和测试对比图');
    + e2 a7 P* w5 D  _
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
    / X8 d5 T* M- n1 H9 a) i
42. legend('网络输出货运量','实际货运量');
    # w# s9 P; x5 G! m
43. xlabel('年份');ylabel('货运量、万吨');\" v# N3 c9 m- Z/ H0 [& D, [; w
    
44. title('运用工具箱货运量学习和测试对比图');  W2 B7 U6 n) |7 Z1 ]! [
    
45. 6 h3 f- i' e1 \
    
46. %新数据仿真, o4 P3 ^8 W( c$ h
    
47. pnew=[73.39,75.55
    % a9 K* a\" j  Z: [0 S0 K# D% F) b
48.      3.9635,4.0975
    * K% P% @; I4 ^9 T) f3 i2 o! Y\" l/ A
49.      0.9880,1.0268];% n, _8 p/ i3 ?1 D& A- o
    
50. pnewn=mapminmax('apply',pnew,ps1);
    2 h/ q* E$ p5 c( x- k- t
51. anewn=sim(net,pnewn);0 ?! P0 s\" p4 @# Z& z
    
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

且生

1. clc
   % g' j% _- B  v& 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,...
   % }8 @6 V4 |  ^- J/ ^! z
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];
     L  g1 H: M, q* t. E6 C
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,...
   6 \0 P7 q) A5 J5 i& L: i: B# U6 w
5.     2.5,2.6,2.7,2.85,2.95,3.10];, o1 T: S5 @\" L% [( j. r
   
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,...
   5 h- P\" E* N\" [* G; Z) ?
7.     0.56,0.59,0.59,0.67,0.69,0.79];% O8 \! [& @& ?* S) [2 ~& L1 B5 Z
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...0 O  O  ^% {( }. c  E$ |# Z5 c
   
9.      22598,25107,33442,36836,40548,42927,43462];) n\" K8 I' |+ g  a: I9 b
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...5 ~2 f; p& P& F1 W0 D$ A, n
    
11.      13320,16762,18673,20724,20803,21804];4 H$ U1 L+ Q7 W! y1 T* I
    
12. p=[sqrs;sqjdcs;sqglmj];
    ; J! _8 l2 [9 Y! A8 L
13. t=[glkyl;glhyl];
    + i( ]/ z  f# f# \$ e$ L. U
14. %数据归一化
    ! \0 o  T( @% B$ f) z8 p
15. [pn,ps1]=mapminmax(p);
    6 H) Q+ k) m* ]4 b: C\" z
16. [tn,ps2]=mapminmax(t);  R1 h4 n* b- Z1 d# c
    
17. %dx=[-1,1;-1,1;-1,1];7 \+ \3 G% `1 L2 W4 G
    
18. [color=Red] p0=minmax(pn);t0=minmax(tn);[/color]2 ]. C2 I3 S# W* y# o
    
19. 
    & y6 R! y  L/ U
20. %BP网络训练
    7 F7 U& t, M& I' z5 f& u( X
21. [color=Red] net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); [/color]: {3 r. N3 J( ~# l  Z
    
22. net.trainParam.show=1000;6 V3 {4 `! X) t% P' r
    
23. net.trainParam.Lr=0.05;% n  b# H( o* D  c
    
24. net.trainParam.epochs=50000;
    ; Y  ^$ s0 p+ e8 g3 x8 F* ~* A% h
25. [color=Red] net.trainParam.goal=0.65*10^(-5);[/color]6 `6 Q- }8 y  o* B
    
26. net=train(net,pn,tn);, A. y3 u* n- K# J
    
27. & v1 C\" U. B' v% v2 X
    
28. %利用原数据对BP网络仿真6 y( }0 ~: n  E) c( T, g- x\" S2 ~. N) L
    
29. an=sim(net,pn);
    + ]. B, }6 m# n1 g8 g2 F( z2 N
30. a=mapminmax('reverse',an,ps2);
    # b# V7 v. W, d4 a' q& J3 |
31. ' [- F' P7 D9 N# N
    
32. %仿真结果与原数据对比测试
    ' a0 ^* H! _; c3 E# i\" ^
33. x=1990:2009;
    1 s( e0 P' \! J+ {5 E& ^; @1 }: L
34. newk=a(1,:);
    \" _: E* }4 t' d# Y
35. newh=a(2,:);
    # \: F/ l  r5 Q
36. figure(2);1 V9 c) d% m1 a- A
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
    1 D4 E4 S\" q6 V+ c
38. legend('网络输出客运量','实际客运量');. j3 f5 t4 ?\" H' ^! n\" e
    
39. xlabel('年份');ylabel('货运量、万人');
    ! S0 z7 Z) Z& D( M% o
40. title('运用工具箱客运量学习和测试对比图');
      k) c( O- ~7 h+ y
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');, d) b% k% f. l8 J3 y5 h8 L  X6 I
    
42. legend('网络输出货运量','实际货运量');2 |9 M- b: E# z% W  s; H
    
43. xlabel('年份');ylabel('货运量、万吨');
    % W9 C% e3 h  R+ K# G; X5 ~3 J# ^
44. title('运用工具箱货运量学习和测试对比图');! Q& R* r' A# w' G0 ^\" W
    
45. 
    \" ~+ i9 P% @' B3 A: J6 b
46. %新数据仿真
    % `7 b* w  g$ s& g
47. pnew=[73.39,75.55
    3 g# n\" M3 o' e. M) h
48.      3.9635,4.0975
    \" P' w9 {% Q! q, w% `# a, Z2 q6 H
49.      0.9880,1.0268];
    \" c' Q% S6 ~0 S+ }: j
50. pnewn=mapminmax('apply',pnew,ps1);
    - G8 V2 A! \& K( p# l) i! U
51. anewn=sim(net,pnewn);4 l1 g: E/ H( a: @$ A\" T% g
    
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

且生

求书名及页码……

T-Eric

gancm 发表于 2014-7-19 12:30
) o& r, [6 Q, a5 w% g) 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 ...

+ P8 w* A( j% V2 o) c9 Z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');
我修改为：
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
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);
; o! N( b/ o7 Ga=postmnmx(an,mint,maxt);
+ Q) I* N2 n1 W3 ]; i5 c/ Jpnewn=tramnmx(pnew,minp,maxp);
anew=postmnmx(anewn,mint,maxt)
: \: y$ F2 w! k* H  N4 `& j我修改为：
7 D5 A: n) B2 M% F$ L& E[pn,ps1]=mapminmax(p);[tn,ps2]=mapminmax(t);
a=mapminmax('reverse',an,ps2);
pnewn=mapminmax('apply',pnew,ps1);
1 |6 k. {8 t. G+ B; aanew=mapminmax('reverse',anewn,ps2)

$ R" L/ f5 ]* y% s8 c+ l原程序为：
% g9 d' v. L8 x( `%原始数据输入
clc
$ z; R% F. o9 t$ Psqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
" v& d+ q( |' q1 i+ e    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,...
- v; v) ?1 Q8 ^5 A4 u    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,...
) ?6 T- e2 ^2 A' z+ e6 f+ C$ d    22598,25107,33442,36836,40548,42927,43462];
- F8 b( P% A" C* p/ j: K( bglhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
7 u7 D+ u# v. o    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
* Z5 U% ]% X. E) j$ G7 f( dt=[glkyl;glhyl];
; |' ?& N- q! S) b
$ H- b4 _; g% S# \9 v2 n" p7 P%数据归一化
% ~) I' \8 J; K) Z' r, E[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
2 h1 q$ i9 @: s0 r  s6 e  zdx=[-1,1;-1,1;-1,1];
( ]; k: T% T; ~( l5 Z/ n
6 r2 H4 z6 [$ r8 D%BP网络训练
net=newff(dx,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
8 t' ^. ~8 b/ r. E: ?1 qnet.trainParam.show=1000;
8 U! L9 @/ c3 ~* H  nnet.trainParam.Lr=0.05;
net.trainParam.epochs=50000;
+ y: E) t) ]1 o; [) jnet.trainParam.goal=0.65*10^(-3);
- ~' Z( D; n5 m; B) znet=train(net,pn,tn);
' K) f! ~1 N3 ?, }6 Z/ G! U
) h! x. y' i4 g$ Y%利用原数据对BP网络仿真
an=sim(net,pn);
a=postmnmx(an,mint,maxt);
3 c6 ^  Y7 Q1 |! F8 j; d
0 c/ T; w6 @; n# f%仿真结果与原数据对比测试
x=1990:2009;
" d% J4 N8 z* V7 ~newk=a(1,;
' [9 y! W* x( S# @# j& inewh=a(2,;
# P. [5 ~+ }0 k) vfigure(2);
& Q+ L, f0 S+ }) R- 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');
xlabel('年份');ylabel('货运量、万人');
* ]: ?+ O; ]" }6 V$ stitle('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--+');
legend('网络输出货运量','实际货运量');
xlabel('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
/ t; j. A5 q( V6 ], X& m( m    3.9635,4.0975
    0.9880,1.0268];
. o, O# `# s+ W2 N( G9 u$ ~. Hpnewn=tramnmx(pnew,minp,maxp);
anewn=sim(net,pnewn);
anew=postmnmx(anewn,mint,maxt)

3 M1 V9 V. l4 m/ I, {& D( \修改后程序为：
%原始数据输入
# |& `: \" f. x. y) L  nclc
, D$ l* v! t2 n8 f' Ksqrs=[20.55,22.44,25.73,27.13,29.45,30.1,30.96,34.06,36.42,38.09,39.13,39.99,...
* m. Q  h7 B7 c& w) y    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,...
6 l: ?, m4 k5 d    2.5,2.6,2.7,2.85,2.95,3.10];
, O7 F5 y$ c& I, I0 F  r% tsqglmj=[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,...
$ _/ L5 R% m+ P- u8 B8 ]3 g; F* K7 p    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,...
+ O8 e" k% b& n) C* R    22598,25107,33442,36836,40548,42927,43462];
0 n. `2 u% U5 ~1 u5 }7 v# V; A2 t1 {glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
$ K% N7 Z6 Z+ g* P    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];

%数据归一化
/ |  }; s+ i! _3 U2 D% S  H[pn,ps1]=mapminmax(p);
+ P+ w0 o/ x3 P. ^2 w/ V2 e" S[tn,ps2]=mapminmax(t);

  d9 x  O# `1 B) Q( L/ x%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
1 {1 H7 {; Y% f8 y; Pnet.trainParam.show=1000;
net.trainParam.Lr=0.05;
3 g, V0 q6 ~7 B: }! T' @! Nnet.trainParam.epochs=50000;
$ ?; r8 Y- n9 a- e+ }$ P6 Mnet.trainParam.goal=0.65*10^(-3);
; A' E5 g( O# L6 G7 R, t  Ynet=train(net,pn,tn);

. S$ \* f7 E2 p) H- R! _%利用原数据对BP网络仿真
6 g! i7 h$ p4 c% @/ Can=sim(net,pn);
a=mapminmax('reverse',an,ps2);

%仿真结果与原数据对比测试
x=1990:2009;
$ h# Z+ ~9 d7 }5 K6 Xnewk=a(1,;
newh=a(2,;
# u/ U$ U8 a' [) P' c7 v, hfigure(2);
* @+ b1 a; X+ e$ ~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');
0 ~* b$ p4 t' W: q: ?' W0 a* ?1 t6 o. Ytitle('运用工具箱客运量学习和测试对比图');
* B( }; j5 B3 Jsubplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
legend('网络输出货运量','实际货运量');
xlabel('年份');ylabel('货运量、万吨');
3 |0 b# |* @( e' D0 T2 N. K* h+ Dtitle('运用工具箱货运量学习和测试对比图');

: {1 o- e. T/ d" W%新数据仿真
2 p( t& {; n5 `, `pnew=[73.39,75.55
" q3 v. C! R4 `! p# V    3.9635,4.0975
    0.9880,1.0268];
0 M  ], V9 b. E& L) Upnewn=mapminmax('apply',pnew,ps1);
anewn=sim(net,pnewn);
" W- \# S6 L; g" ]: R" Sanew=mapminmax('reverse',anewn,ps2)
（修改的地方用颜色标记了）
4 e8 a# x9 [& z% {; f) ]麻烦您帮忙指出其中的问题，万分感谢！