摘要
针对线控转向汽车的可靠性和安全性以及故障诊断方法的不足,提出了一种基于软计算的汽车线控转向故障诊断方法,该方法利用软计算中的粗糙集和粒子群优化的径向基神经网络进行结合。将粗糙集作为径向基神经网络的输入处理,对样本数据进行属性约简,约简后的属性集作为径向基神经网络的输入以达到缩短网络训练时间的目的。采用粒子群算法对径向基神经网络的基函数中心值和宽度进行编码和寻优,并使用得到的最优中心值和宽度组建径向基神经网络,使得径向基神经网络的样本训练误差相比未优化之前有一定程度的降低。然后使用训练好的神经网络对故障样本进行测试,测试结果表明,该方法加快了神经网络的训练速度,提高了神经网络的诊断准确度。
In view of the reliability, and safety problems and the imperfectness of fault diagnostic approach for automobile steer-by-wire system, a method based on soft computing is designed, rough set （RS） combined with radical basis function （RBF） neural network op- timized by particle swarm optimization （PSO） is presented in this method. A rough set model is utilized as an input processor to reduce the redundant information of the samples. On the basis of the reduction, the reduced attributes can be extracted. The central position and the width of the basis function in the RBF network is encoded and optimized by the PSO algorithm. Then a simulation is executed in the trained network. Higher training Speed and diagnostic accuracy of the network are shown in the result.
出处
《控制工程》
CSCD
北大核心
2012年第2期316-319,共4页
Control Engineering of China
基金
广西研究生教育创新计划基金项目(2011105940811M01)
关键词
软计算
线控转向系统
故障诊断
粗糙集
径向基神经网络
粒子群优化
soft computing
steer-by-wire system
fault diagnosis
rough set
radical basis function neural network
particle swarm op-timization
