基于IPSO-PNN的电动汽车故障诊断

为了提高电动汽车故障诊断的准确性,提出了一种基于改进粒子群(Improved Particle Swarm Optimization,IPSO)算法优化概率神经网络(Probabilistic Neural Network,PNN)的电动汽车故障诊断方法,即基于IPSO-PNN的电动汽车故障诊断方法。首先,研究了基于PNN的电动汽车故障诊断模型,分析了PNN的平滑因子对该模型诊断准确率的影响;其次,在粒子群(Particle Swarm Optimization,PSO)算法中引入频率粒子群和采用动态惯性权重,改善PSO算法的全局和局部寻优能力,利用IPSO算法优化基于PNN的电动汽车故障诊断模型的平滑因子,以改善模型的分类能力;最后进行仿真与分析。仿真结果表明:相较于基于PSO-PNN的电动汽车故障诊断方法,基于IPSO-PNN的电动汽车故障诊断方法具有更高的诊断准确率和诊断速度。

Electric vehicle fault diagnosis based on IPSO-PNN

In order to improve the accuracy of electric vehicle fault diagnosis,an electric vehicle fault diagnosis method based on Improved Particle Swarm Optimization(IPSO)algorithm optimized Probabilistic Neural Network(PNN)was adopted.That is an electric vehicle fault diagnosis method based on IPSO-PNN.Firstly,the fault diagnosis model of electric vehicle based on PNN was studied,and the influence of the smoothing factor of PNN on the accuracy of model diagnosis was analyzed.Secondly,using frequency particle swarm was introduced and the dynamic inertia weights was used in Particle Swarm Optimization(PSO)algorithm to improve the global and local optimization capabilities of PSO algorithm,IPSO algorithm was used to optimize the smoothing factor besed on PNN electric vehicle fault diagnosis model.Finally,the simulation and analysis were carried out.Simulation results show that,compared with the electric vehicle fault diagnosis model based on PSO-PNN,the diagnosis model based on IPSO-PNN has higher diagnostic accuracy and speed.