神经网络的样本值对整个网络能否正常运行及误差的大小非常重要，因此本程序不采用让用户自己输入样本植的方法，而自动设定了样本值

该网络采用的作用函数为Sigmoid函数，即f(x)=1/(1+e-x)。网络先对输入的样本值进行训练，当达到规定的训练次数或者网络的整体误差达到规定的精确度时就停止训练。此后，可以输入成绩运行网络得出运行结果。

-neural network for the entire sample values of normal network operation and the size of error is very important. So do not use this procedure to allow users to import their own samples of the plants, and the sample is automatically set the value of the network to function, Sigmoid function, that f (x) = 1 / (an e-x). The network of input sample values training When the required number of training or network error to the overall accuracy of the time to stop training. Since then, the results can be imported to come running network operating results.