在实际应用中，要根据具体的使用领域和使用需求，选取适当规模和领域的训练数据对BP 神经网络进行训练，从而不断的调整各个隐含层神经元的权系数和阈值范围，从而能够使最后获取的实际输出在所期望的误差范围之内。

-In practical applications, according to the specific field of use and usage requirements, select the appropriate scale and scope of training data on BP neural network training, so as to continuously adjust the weights and threshold values ​ ​ for each hidden layer neurons, thereby enabling Being within the actual output of the last error in the desired range.