人工神经网络(Artificial Neural Networks)是模仿人脑结构和功能的一种信息处理系统。该方法在处理非线性问题上具有其独特的优越性,在工程应用中越来越被人们所关注。与此同时,神经网络计算的精度和收敛速度也成为人们普遍关心的问题。本文试验发现BP算法在归一化过程中存在最值误差和区间变形误差,在此基础上,本文对常规BP网络进行了改进,提高了BP网络的预测精度。 Artificial Neural Networks is an information processing system that mimics the structure and function of the human brain. This method has its unique superiority in dealing with nonlinear problems and has drawn more and more attention in engineering applications. At the same time, the accuracy and convergence speed of neural network computing have also become a common concern. In this paper, we find that the BP algorithm has the most error and interval distortion in the normalization process. On this basis, this paper improves the conventional BP network and improves the prediction accuracy of the BP network.