基于粒计算-神经网络的汽车电子控制系统故障分析  

Fault Analysis of Automotive Electronic Control System Based on Granular Computation Neural Network

作　　者：刘涛[1] LIU Tao(Anhui Vocational College of Electronics&Information Technology,Bengbu Anhui 233040,China)

机构地区：[1]安徽电子信息职业技术学院,安徽蚌埠233040

出　　处：《佳木斯大学学报（自然科学版）》2021年第4期59-63,67,共6页Journal of Jiamusi University：Natural Science Edition

基　　金：安徽省教育厅质量工程建设(2019xqsxzx57)。

摘　　要：汽车电子控制系统故障数据通常具有规模庞大、非线性等特点,相关的故障分析技术成为相关学者共同关注的话题。此次研究利用粒计算-神经网络诊断汽车电子控制系统故障问题,分析粒计算的知识约简算法、BP神经网络以及RBF神经网络的前提下,构建面向粒计算-神经网络的汽车电子控制系统故障方法。研究结果显示,BP和GrC-BP神经网络在训练步长91次和28次满足训练要求,相应的MSE分别为0.0013和0.0018。四种神经网络对汽车电子控制系统的故障诊断误差进行了比较分析。The fault data of automotive electronic control system usually has the characteristics of large scale and non-linear,and the related fault analysis technology has become a common topic of concern of relevant scholars.This research uses granular computation-neural network to diagnose the fault of automobile electronic control system,analyzes the knowledge reduction algorithm of granular computing,BP neural network and RBF neural network,and constructs the fault method of automobile electronic control system oriented togranular computation-neural network.The results show that BP and GRC BP neural networks meet the training requirements in 91 and 28 training steps,and the corresponding MSE are 0.0013 and 0.0018,respectively.This paper also compares and analyzes the fault diagnosis errors of four kinds of neural networks for automotive electronic control system.

关 键 词：粒计算 神经网络 故障分析 电子控制系统 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]