基于扩展T-S模型的PSO神经网络在故障诊断中的应用  

Application of PSO Neural Network Based on Extended T-S Model in Fault Diagnosis

作　　者：王建芳[1] 李伟华[1] 

机构地区：[1]西北工业大学计算机学院,西安710072

出　　处：《计算机科学》2009年第9期224-226,245,共4页Computer Science

基　　金：部委专项(51315080404);武器装备预研基金项目(9140A17050206HK03);航空科技创新基金(08E53003)资助

摘　　要：针对现实故障现象具有模糊性和非线性的特点,提出了一种利用自适应扩展T-S(Takagi-Sugeno)模糊模型的PSO(Particle Swarm Optimization)算法和神经网络相结合的新型智能结构化算法来进行故障诊断的新方法。首先通过自适应的高斯函数来更改基本T-S模糊模型中的隶属度函数,进而使用扩展的T-S模糊模型来调整PSO算法的参数。然后使用该PSO算法作为神经网络的学习训练算法来进行训练。最后将此算法用于齿轮箱实测故障诊断。诊断结果显示均方误差提高了0.1981%。通过不同模型的诊断结果比较,表明本方法便捷、高效,为解决故障诊断问题提供了一条新途径。To solve fuzzy and non-linear features of faults, a fault diagnosis method was developed based on extended T- S （Takagi-Sugeno） fuzzy model of self-adaptive disturbed PSO （Particle Swarm Optimization） combined with Neural Network. Firstly, the membership function of the basic T-S fuzzy model was modified by the adaptive gaussian function, and then the extended T-S model was used to adjust the PSO parameter. Secondly, the neural network was trained by the modified PSO algorithm. Finally, the proposed method in the paper was applied to fault diagnosis of gear-box. The diagnosis results show that the mean square error is improved 0. 1981% ,meanwhile,comparisons with the diagnosis result of the different models show the method in the paper is convenient, efficient, and provides a new approach to fault diagnosis.

关 键 词：模糊模型 离子群优化算法 神经网络 故障诊断 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程]