摘要
针对典型BP神经网络在装甲车辆电气系统电路板故障诊断中容易出现自适应效果差、局部极小值等问题,通过引入遗传算法(genetic algorithm,GA),对典型BP神经网络的各层参数进行优化,从而对典型BP神经网络故障诊断模型进行改进。为了验证模型的可靠性,以装甲车辆电气系统中80式灭火系统控制盒电路板故障诊断数据为例,对参数优化后的模型进行分析验证,结果表明,改进后的模型能够有效克服BP神经网络模型自适应不够的问题,并避免网络陷入局部极小值,从而有效提升装甲车辆电气系统电路板故障诊断效率和质量。
Aiming at the problems of poor adaptive effect and local minimum of typical BP neural network in circuit board fault diagnosis of armored vehicle electrical system,the parameters of each layer of typical BP neural network are optimized by introducing genetic algorithm(GA),so as to improve the fault diagnosis model of typical BP neural network.In order to verify the reliability of the model,taking the circuit board fault diagnosis data of the control box of type 80 fire extinguishing system in the electrical system of armored vehicles as an example,the model after parameter optimization is analyzed and verified.The results show that the improved model can effectively overcome the problem of insufficient adaptability of BP neural network model and avoid the network falling into local minimum,So as to effectively improve the efficiency and quality of fault diagnosis of circuit board of armored vehicle electrical system.
作者
谢永成
李光升
魏宁
李刚
XIE Yong-cheng;LI Guang-sheng;WEI Ning;LI Gang(Department of Weapons and Control,Army Armored Forces Academy,Beijing 100072,China)
出处
《自动化与仪表》
2022年第8期97-101,共5页
Automation & Instrumentation
基金
陆军装备部重点项目(LJ20191A050223)。
关键词
遗传算法
BP神经网络
装甲车辆
电气系统
故障诊断
genetic algorithm(GA)
BP neural network
armored vehicle
electrical system
fault diagnosis
作者简介
谢永成(1963—),男,博士,教授,研究方向为装甲车辆电气系统故障诊断;李光升(1972—),男,硕士,副教授,研究方向为装甲车辆电气系统故障诊断。
