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！！

%原始数据输入
clc
/ F9 C( |0 T$ T/ }0 G- ysqrs=[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];
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 _! v0 V8 `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];
4 S8 g& @& o: q6 J+ p# Kglkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
2 @, x1 V* S( h3 ~2 k. A    22598,25107,33442,36836,40548,42927,43462];
glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
9 w7 E0 l$ O! y* R! Q' `3 j    13320,16762,18673,20724,20803,21804];
. Z8 E8 i8 p( T; Op=[sqrs;sqjdcs;sqglmj];
: o# s: X. i/ c3 Q8 v, b4 Et=[glkyl;glhyl];

%数据归一化
[pn,ps1]=mapminmax(p);
$ D$ n: }8 S9 j[tn,ps2]=mapminmax(t);
dx=[-1,1;-1,1;-1,1];
$ ^3 D4 K: V+ g
# G$ R; C; K% t%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
net.trainParam.show=1000;
; f1 U3 K3 `  M3 |% ~* m* e( Hnet.trainParam.Lr=0.05;
net.trainParam.epochs=50000;
* I  i8 b; U( G8 G. C/ U3 `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);
: F' k' g) E4 ~7 e4 s/ J$ {* oa=mapminmax('reverse',an,ps2);
7 `: `. J& r2 U; L6 f4 Y/ v  X
%仿真结果与原数据对比测试
x=1990:2009;
; ~/ b$ O& ^) p/ k% R% L& jnewk=a(1,:);
5 L! u) G+ Y1 M- y! dnewh=a(2,:);
- ~4 S" x- H- h& V+ vfigure(2);
subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
3 o; s8 _+ W+ X7 q- m/ U) s' Xlegend('网络输出客运量','实际客运量');
+ J- m& X6 ]% R$ dxlabel('年份');ylabel('货运量、万人');
: C& \6 [# e" B! B0 Mtitle('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');
% \) I2 M1 D) txlabel('年份');ylabel('货运量、万吨');
9 i7 ?3 E/ Z* X% [( Q; jtitle('运用工具箱货运量学习和测试对比图');
/ X3 t, s) B- q8 ]9 F
%新数据仿真
% P4 g" `/ p( rpnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
+ ^1 _2 f7 y, u) v1 n' {& E& mpnewn=mapminmax('apply',pnew,ps1);
anewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)

; O* t( Z' ^' \

T-Eric

狼之魂汪洋 发表于 2014-8-6 14:00
3 D; K6 U: O& }& p你这例题是在什么书上找的？

( ?  r* `! h# t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
   6 E- [7 H7 o) p2 `/ t. @
2. net=newff(p0,t0,6,{'tansig'},'traingd');
   ; S( ?- c1 t# ^
3. net.trainParam.show=1000;3 x% T7 |# H4 }
   
4. net.trainParam.Lr=0.05;$ d$ _. v- S7 l
   
5. net.trainParam.epochs=2000;$ `& U: J$ ?4 |) O+ ^0 J
   
6. net.trainParam.goal=0.65*10^(-4);
   % [5 Y* ~0 f3 Q8 O, \
7. net=train(net,pn,tn);

复制代码

我只改了这里面的，
# [9 S& R5 S( H5 R" `一、改成单隐含层的，6个节点
) j+ ?( p* Q5 k$ d- X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。
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
+ C: T; }6 T% b2 f' ^6 w( K" t9 G2 j忽略我上面那个，改动的地方在18，21，25行……求批

, `; [' ?4 j8 v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' P7 x. m4 j' ^; ]\" E
   
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( |- s9 U8 f# @/ h0 E+ w
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];9 n& N) ~8 j3 I
   
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 U8 y6 J8 D0 b, `2 f
5.     2.5,2.6,2.7,2.85,2.95,3.10];; y* J: M  T- `* b. v
   
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,...& B. K& ?8 I1 j& N2 p9 m2 v, W
   
7.     0.56,0.59,0.59,0.67,0.69,0.79];& W) L1 [, t- H6 z7 `! `/ q
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...
   5 [/ g, Z! E. z+ u: W+ ]* H: @6 ~' R
9.      22598,25107,33442,36836,40548,42927,43462];
   0 P) L- N, k, K' s8 O
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...$ |7 C. j# d5 M+ b: b1 e
    
11.      13320,16762,18673,20724,20803,21804];
    $ c9 s' ~4 R, v5 T( P% w8 f
12. p=[sqrs;sqjdcs;sqglmj];
    8 C; K3 m0 c( K; z& k+ ?
13. t=[glkyl;glhyl];
    + @\" a1 [3 C! x5 q\" t' r
14. %数据归一化  `1 O1 w0 H8 {8 S* S* o0 l* W
    
15. [pn,ps1]=mapminmax(p);: i# L2 z- R4 _. u& E5 h6 U7 \, ]
    
16. [tn,ps2]=mapminmax(t);
    + h2 w  W\" B* S$ \\" f6 _
17. %dx=[-1,1;-1,1;-1,1];: z) ]4 g) h# \/ o
    
18. p0=minmax(pn);t0=minmax(tn);$ d2 V' k0 z/ U4 P# ?( V6 E
    
19. / G4 ~6 Y2 ]4 B$ ?- ]# Z7 n
    
20. %BP网络训练) q% S9 \/ k. j8 S, S9 B6 \  o
    
21. net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
    7 B$ h5 b# k+ }9 X$ n8 O) v; M8 n' w
22. net.trainParam.show=1000;
    * F  M, ]3 z) D5 [9 D$ {
23. net.trainParam.Lr=0.05;& i8 ?3 _( Z4 ]5 T. B+ X* e
    
24. net.trainParam.epochs=50000;
    ) G* ^; u& t) X( L
25. net.trainParam.goal=0.65*10^(-5);3 S8 g4 j1 b, q6 I8 y  v\" j7 G
    
26. net=train(net,pn,tn);
    8 ^# e# d5 X  H* I' t: l8 g
27. ( _* Y* \9 [3 z- H  R
    
28. %利用原数据对BP网络仿真
    ; @2 G/ U1 J* z( }5 T/ z
29. an=sim(net,pn);& z+ a1 _4 ~$ l+ P
    
30. a=mapminmax('reverse',an,ps2);
    1 ~8 E2 y/ K8 |0 M\" S, ~
31. 
    ) v3 ?0 e& L6 m% I7 D
32. %仿真结果与原数据对比测试
    ! Z; q1 }. W8 X\" M0 I2 d3 `
33. x=1990:2009;
    - v# Z4 B9 o' Y: x
34. newk=a(1,:);/ n$ _9 H+ i6 v0 \
    
35. newh=a(2,:);
    % Z& F' i! s; |  J! ~
36. figure(2);- a/ U% j# v7 M/ ]. U' v+ O/ W6 t& q8 P
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
    $ g& d0 Z& r5 Q* T- V- `0 k# T4 `' U
38. legend('网络输出客运量','实际客运量');* z7 n9 x' n% i% d7 @7 `1 H
    
39. xlabel('年份');ylabel('货运量、万人');
    / ^/ h7 n, T/ W& u
40. title('运用工具箱客运量学习和测试对比图');
    5 j' \7 R7 |& T6 w+ g* S1 s3 x- a
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');4 x& \- d' ]# W5 {+ T' G4 w3 u: N+ D
    
42. legend('网络输出货运量','实际货运量');
    # x! w- v' ?( R7 s5 H2 [3 O, E
43. xlabel('年份');ylabel('货运量、万吨');) Y- j) I- A% Z) V$ c$ h
    
44. title('运用工具箱货运量学习和测试对比图');
    + h! F3 U3 i* b: v\" R4 A
45. 3 U# w; |\" Z( `: K) U3 ?
    
46. %新数据仿真
    . s! X0 n, D# m' K2 _, L* p  W' f
47. pnew=[73.39,75.553 ]( ]; W- u8 ^1 k
    
48.      3.9635,4.09753 [# Y! A: `% [- [3 V6 H8 ]3 `
    
49.      0.9880,1.0268];
    & U. j0 M4 s2 N: N& ]  g$ T, E6 R
50. pnewn=mapminmax('apply',pnew,ps1);
    & A4 [  C+ {5 g8 |( z
51. anewn=sim(net,pnewn);9 e/ i8 S) D$ o, o2 k
    
52. anew=mapminmax('reverse',anewn,ps2)

复制代码

且生

1. clc
   9 ~3 S0 I* u2 }) O4 j* 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,...
   - y6 [. Y( ?% }9 K0 e, Q. F
3.     41.93,44.59,47.30,52.89,55.73,56.76,59.17,60.63];% @( l8 }- O, K9 L, i
   
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,...) Z. ~0 J5 M7 ^- i, w# U
   
5.     2.5,2.6,2.7,2.85,2.95,3.10];2 n: T' ~\" g2 y% N& S& |# t; l
   
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,...$ l( t. \2 E0 z\" l) u0 D, w( a
   
7.     0.56,0.59,0.59,0.67,0.69,0.79];$ w3 D, t6 \& o) ?* x
   
8. glkyl=[5126,6217,7730,9145,10460,11387,12353,15750,18304,19836,21024,19490,20433,...5 m7 R% a# a' z' M0 w
   
9.      22598,25107,33442,36836,40548,42927,43462];! f9 n1 N7 t; ^- Y! F8 u9 o
   
10. glhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
    7 y: }3 {# E6 I% A+ d7 u
11.      13320,16762,18673,20724,20803,21804];
    ; O+ n) P3 Y' {
12. p=[sqrs;sqjdcs;sqglmj];
    # p7 W  s0 R$ Q( B) n, X+ X
13. t=[glkyl;glhyl];
    7 i7 |% D9 |; M/ R\" x8 O6 I! @
14. %数据归一化
    8 Q5 F8 |7 b- B( Z0 X1 Y
15. [pn,ps1]=mapminmax(p);
    ! O) A( e- [! T/ B\" x
16. [tn,ps2]=mapminmax(t);
    ; a' |! M# |% g
17. %dx=[-1,1;-1,1;-1,1];6 k2 a) J. X; I* o, U0 W, x
    
18. [color=Red] p0=minmax(pn);t0=minmax(tn);[/color]; {3 \0 L& E% a! O- [! Z# ^9 N
    
19. 
    & j& t# S9 i! v; x9 ~+ p
20. %BP网络训练8 F\" C1 N, r- f! ~  x- w
    
21. [color=Red] net=newff(p0,t0,[3,7,2],{'tansig','tansig','purelin'},'traingdx'); [/color]9 A7 s+ ^- I7 e\" ~; J/ A
    
22. net.trainParam.show=1000;& G. c\" g  f0 a' [) g% W
    
23. net.trainParam.Lr=0.05;
    6 r0 S* Z, o7 q0 q/ T7 }1 \
24. net.trainParam.epochs=50000;7 g2 L3 ^- H0 K& a
    
25. [color=Red] net.trainParam.goal=0.65*10^(-5);[/color]
    4 [5 |: C' K  B; l  @* v
26. net=train(net,pn,tn);9 U$ q' e# s- u
    
27. 
    : G2 _$ u& J9 Z3 s\" a
28. %利用原数据对BP网络仿真0 q$ E* U6 g4 F& ?& z
    
29. an=sim(net,pn);3 Q6 T: r# e  j2 j
    
30. a=mapminmax('reverse',an,ps2);
    ( n- Y; u6 o4 R
31. 
    \" @  ]6 N7 n+ O9 X- n
32. %仿真结果与原数据对比测试 7 j& J  M$ f$ j\" |8 \* ?
    
33. x=1990:2009;  a7 |3 w\" h; O5 ~
    
34. newk=a(1,:);8 f) D6 M3 {! J  n3 u' i* g, h* D* V
    
35. newh=a(2,:);9 M  p. Q( ^$ C/ O8 L/ w5 E& J
    
36. figure(2);6 d' m9 d8 z/ D2 b0 O
    
37. subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');: S: N  ?& M' g
    
38. legend('网络输出客运量','实际客运量');/ Q; ^) V7 p3 @; @
    
39. xlabel('年份');ylabel('货运量、万人');4 Z* r. \' q' F  G
    
40. title('运用工具箱客运量学习和测试对比图');2 M1 k6 {0 C8 }% l! e
    
41. subplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
    4 Z! `* b2 b2 \4 p/ T3 [# r
42. legend('网络输出货运量','实际货运量');
    ; P3 a5 C8 F  p/ `# {: d( O
43. xlabel('年份');ylabel('货运量、万吨');
    \" @1 W; k  g  q/ O+ m
44. title('运用工具箱货运量学习和测试对比图');& s) o) m& ^5 y1 \0 I( \+ [, o5 ?/ B
    
45. & k9 U' o  q2 u  M2 P
    
46. %新数据仿真
    ' w4 H0 [! y, M; ^! _
47. pnew=[73.39,75.55
    % H* h' p\" [# ?$ T
48.      3.9635,4.09757 _  o+ l* @. ~! n
    
49.      0.9880,1.0268];
    5 o3 g6 t  ]) a\" l7 o7 R7 g5 s% U  Q. F
50. pnewn=mapminmax('apply',pnew,ps1);
    ! ]- u3 Z1 D. ?5 v+ }# W
51. anewn=sim(net,pnewn);
    + ]6 R9 s( N* b! @: {7 X% b
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
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 ...

因为版本更新，Matlab中的Newff命令用法有所改变，原命令为：
% r( V2 n4 F3 J7 X4 _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);
a=postmnmx(an,mint,maxt);
- ^$ E4 }  D: H9 [* cpnewn=tramnmx(pnew,minp,maxp);
anew=postmnmx(anewn,mint,maxt)
我修改为：
1 v0 l' T2 }1 G# C[pn,ps1]=mapminmax(p);[tn,ps2]=mapminmax(t);
& v1 L9 m# ^( a2 A: Ba=mapminmax('reverse',an,ps2);
, G6 @8 g8 C7 _pnewn=mapminmax('apply',pnew,ps1);
anew=mapminmax('reverse',anewn,ps2)

原程序为：
%原始数据输入
clc
7 N1 V: k- _! }* _1 ]4 \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 `  R7 W# m2 z1 hsqjdcs=[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,...
# h" W/ [! ~# l9 z* j. n    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,...
    22598,25107,33442,36836,40548,42927,43462];
0 U1 n8 I" D/ P# b1 ?; S  gglhyl=[1237,1379,1385,1399,1663,1714,1834,4322,8132,8936,11099,11203,10524,11115,...
% g" F! @, z) Z+ u    13320,16762,18673,20724,20803,21804];
p=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];
1 O* [" n; u, J# y
: G: ~( O% R* `%数据归一化
! x/ }( T& v  V1 u, N; X# L# O5 t[pn,minp,maxp,tn,mint,maxt]=premnmx(p,t);
+ G8 _( E7 e  M8 i+ F3 y+ C9 I" {dx=[-1,1;-1,1;-1,1];
* Q1 y7 r, `3 ]1 @
( P, U0 E3 P, c%BP网络训练
net=newff(dx,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
% }: S$ U5 A" p6 ~* G/ x6 v8 lnet.trainParam.show=1000;
. O/ Q) r5 C* U* znet.trainParam.Lr=0.05;
; w$ i/ y# s% H' d4 H+ ]$ wnet.trainParam.epochs=50000;
: G$ t# j; U3 `/ o9 bnet.trainParam.goal=0.65*10^(-3);
net=train(net,pn,tn);

1 B' C. q0 y  ~5 J0 [%利用原数据对BP网络仿真
; N6 p6 k4 j5 W1 [! Jan=sim(net,pn);
/ N3 {7 ]/ ?/ R8 Va=postmnmx(an,mint,maxt);

, p" @- D( `4 ?% D. |%仿真结果与原数据对比测试
6 J% D2 E, l. a" ^& p( y* Dx=1990:2009;
7 y) N% D3 ?9 G$ Y8 V; X; Anewk=a(1,;
newh=a(2,;
: K) [5 ^* M0 T% O2 u7 Dfigure(2);
5 {" M0 k; \( x' Jsubplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
legend('网络输出客运量','实际客运量');
/ I) y* m) S5 f9 ^' s0 B% kxlabel('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--+');
) _; E3 \6 k9 _8 B4 A6 Mlegend('网络输出货运量','实际货运量');
8 k. Y+ F( ~/ z4 f5 r8 H5 b& n$ kxlabel('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');
* F4 C% b, j' B& t- s
6 P$ l& Z6 H, \6 e%新数据仿真
pnew=[73.39,75.55
    3.9635,4.0975
    0.9880,1.0268];
2 _1 Q' U4 K* {+ Gpnewn=tramnmx(pnew,minp,maxp);
anewn=sim(net,pnewn);
% h0 j$ o9 c% L8 {" c/ K% Hanew=postmnmx(anewn,mint,maxt)

- E. S1 `% v- K8 A# m! _2 X修改后程序为：
/ k4 f9 _0 _' W& C( m; Z%原始数据输入
- y( t& e6 l. v6 jclc
" I1 @6 b4 Y% dsqrs=[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];
6 ~& N  c% g! ?, msqjdcs=[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; ^1 F  j    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,...
    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];
5 a8 w( i9 a8 h* _6 ^6 d( Qp=[sqrs;sqjdcs;sqglmj];
t=[glkyl;glhyl];

, e5 a' X2 h  ^5 n%数据归一化
- m8 w% d0 a1 k- \$ g[pn,ps1]=mapminmax(p);
4 \$ E. t8 @" f[tn,ps2]=mapminmax(t);
7 G2 |# f5 k  ~7 P
+ Q- r5 Q% s% w4 M5 X! l( v& }%BP网络训练
net=newff(p,t,[3,7,2],{'tansig','tansig','purelin'},'traingdx');
) b: U9 _7 J4 Wnet.trainParam.show=1000;
net.trainParam.Lr=0.05;
# o# K3 U5 W" d& U; ^" Bnet.trainParam.epochs=50000;
- p: L& n9 c- [. e& l! r/ }net.trainParam.goal=0.65*10^(-3);
0 j! f1 v+ Y" ]! y" nnet=train(net,pn,tn);
4 m+ @( p/ y* j+ c
, Q+ \$ ~1 r8 M+ t$ S1 a6 j  g%利用原数据对BP网络仿真
" V5 o# Z( Q  F7 o- ?, ]an=sim(net,pn);
a=mapminmax('reverse',an,ps2);

+ p/ E; G0 p5 n%仿真结果与原数据对比测试
7 _( L+ g+ g$ y5 V* Hx=1990:2009;
$ E4 y  a; c2 E. \  ~$ C2 Tnewk=a(1,;
# i" O4 A+ P2 g6 w. B+ I! z& Anewh=a(2,;
+ k+ F0 H# t. J, Mfigure(2);
subplot(2,1,1);plot(x,newk,'r-o',x,glkyl,'b--+');
6 _2 F5 R9 Y& d* R6 N( Zlegend('网络输出客运量','实际客运量');
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');
: n3 e  O9 b  O- ~  Ssubplot(2,1,2);plot(x,newh,'r-o',x,glhyl,'b--+');
$ R* I  J3 C9 W. olegend('网络输出货运量','实际货运量');
xlabel('年份');ylabel('货运量、万吨');
# z7 T! R: I3 S' Ztitle('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
: o4 M- k7 [! l) [6 e    3.9635,4.0975
    0.9880,1.0268];
pnewn=mapminmax('apply',pnew,ps1);
) \0 a) l/ V/ m. y! h: oanewn=sim(net,pnewn);
anew=mapminmax('reverse',anewn,ps2)
, S  H9 `( t& s3 u+ E2 S（修改的地方用颜色标记了）
/ D5 M+ t0 H; J. m麻烦您帮忙指出其中的问题，万分感谢！