用BP网络完成函数的逼近 P网络通常有一个或多个隐层，隐层中的神经元均采用sigmoid型变换函数，输出层的神经元采用纯线性变换函数。本例应用一个两层BP网络来完成函数逼近的任务，其中隐层的神经元个数是5。-network to be completed by BP function approximation P networks usually have one or more hidden layer, a hidden layer neurons are used sigmoid-type transformation function, the output layer neurons using pure linear transformation function. The application of a two-tier cases BP network function approximation to complete the task, hidden layer neuron number is 5.