模型误差补偿的神经网络方法研究

Study on the method for compensating model error based on neural networks

本文首先介绍了模型误差补偿的传统方法,附加系统参数法和最小二乘配置法。然后,介绍了神经网络BP算法以及模型误差补偿的神经网络模型结构。论述了神经网络BP改进算法(简称H-BP算法)的思想及其补偿模型误差的原理,并列出了H-BP算法的具体计算步骤。结合一个工程实例,与附加系统参数法和最小二乘配置法等传统补偿方法进行了比较,模型误差补偿的神经网络方法效果更好一些。最后,得出了一些有益的结论。

At first,two traditional methods for compensating function model errors,the method of adding systematic parameters and the least-squares collection method,were introduced in the paper.Second,the BP algorithm of neural network was introduced briefly.A neural network based method for compensating model error was discussed.The special structure of BP network,its calculation steps and the principle of this method were introduced in detail.Then in one engineering project,the results of different methods for compensating model errors were compared with each other.It was shown that the proposed method based on neural network for compensating model error was more effective than traditional methods.At last,some conclusions were reached.