我的神经网络结果为什么都为1,还有对降水量数据预测的时候输入层和输出层应该怎么定?

题目

我的神经网络结果为什么都为1,还有对降水量数据预测的时候输入层和输出层应该怎么定?
clear,clc
T1=[70.67 65.49 61.22;
65.49 61.22 59.53;
61.22 59.53 73.82 ;
59.53 73.82 53.79;
73.82 53.79 62.86;
53.79 62.86 50.94;
62.86 50.94 57.25;
50.94 57.25 53.89;
57.25 53.89 46.91;
53.89 46.91 70.62;
46.91 70.62 61.25;
70.62 61.25 57.05;
61.25 57.05 58.92;
57.05 58.92 57.94;
58.92 57.94 49.32;
57.94 49.32 45.6;
49.32 45.60 43.74;
45.60 43.74 53.83;
43.74 53.83 53.20;
53.83 53.20 55.27;
53.20 55.27 55.34;
55.27 55.34 56.50;
55.34 56.50 55.46;
56.50 55.46 59.47;
55.46 59.47 75.61;
59.47 75.61 66.63;
75.61 66.63 50.20;
66.63 50.20 56.10;
50.20 56.10 62.15;
56.10 62.15 57.76;
62.15 57.76 52.37;
57.76 52.37 52.38;
52.37 52.38 63.80;
52.38 63.80 78.38;
63.80 78.38 56.20;
78.38 56.20 46.86;
56.20 46.86 62.22;
46.86 62.22 50.72;
62.22 50.72 61.15;
50.72 61.15 52.95;
61.15 52.95 56.72;
52.95 56.72 51.12;
56.72 51.12 56.94;
51.12 56.94 64.39;
56.94 64.39 54.77;
64.39 54.77 61.67;
54.77 61.67 51.91;
61.67 51.91 99.27]';%把每三年的数据形成一个输入
T2=[61.22 59.53 73.82 53.79 62.86 50.94 57.25 53.89 46.91 70.62 61.25 57.05 58.92 57.94 49.32 45.6 43.74 53.83 53.2 55.27 55.34 56.5 55.46 59.47 75.61 66.63 50.2 56.1 62.15 57.76 53.37 52.38 63.80 78.38 56.2 46.86 62.22 50.72 61.15 52.95 56.72 51.12 56.94 64.39 54.77 61.67 51.91 99.27];%第四年的组成输出
net=newff(minmax(T1),[5,1],{'tansig','logsig'},'traingdx');
net.trainparam.show=50; %每次循环50次
net.trainParam.epochs=8000;
net.trainParam.goal=0.00001;
net=train(net,T1,T2); %这步是训练网络
T3=sim(net,T1)
计算结果都为1,为什么（我是把前三个数据作为输入层,后一个为输出层,以此类推）

答案

看上去没什么问题,建议先把数据归一化,这样比较好.