基于混合粒子滤波的故障诊断方法①.PDFVIP

计 算 机 系 统 应 用 2012 年 第 2 1 卷 第 12 期 基于混合粒子滤波的故障诊断方法① 董 欣 (渤海大学 管理学院, 锦州 121000) 摘 要: 结合了模糊递归神经网络和粒子群算法, 在此基础上改进粒子滤波的故障诊断方法. 故障诊断作为智能 控制的研究热点, 其算法层出不穷. 粒子滤波故障诊断作为基于解析模型的状态估计诊断方法的一种, 一直在工 业生产中起着重要的作用, 但其自身的缺点也限制了它的进一步发展, 保留粒子滤波的优势, 提出一种基于混合 算法的粒子滤波故障诊断方法, 该方法不但一定程度上解决了权值退化的问题, 而且改进了粒子群算法, 并结合 神经网络算法在啤酒发酵温控系统中形成了故障检测、预测和辨识为一体的故障诊断方法. 关键词: 粒子滤波; 粒子群优化算法; 免疫原理; 故障诊断; 神经网络 Fault Diagnosis Method Based on Hy- PF DONG Xin (School of Management, Bohai University, Jinzhou 121000, China) Abstract: A novel fault diagnosis method is proposed in this paper which is improved with DRFNN and PSO on basis of PF. There are much more algorithms about fault diagnosis because it has become the spot in intelligent control. As one of status estimation diagnostic methods based on the analytical model, particle filter fault diagnosis has been playing an important role in industrial production. However, it is limited by its shortcoming for its further development, a new particle filter fault diagnosis method based on hybrid algorithm is proposed in this paper, which retains the advantages of the particle filter, and it not only solved the problem of the weight degradation to a certain extent, but also improved particle swarm optimization. A system includes fault detection, prediction and recognized is realized with neural network in the beer fermentation temperature control system. Key words: PF; PSO; immune principle; fault diagnosis; neural network 故障诊断(fault diagnosis)是工业生产制造中非常 方法是利用系统的精确的数学模型和可观测输入输出 重要的一个环节, 对于故障诊断的传统分类方法是基 量构造残差信号, 表示实际值和模型期望值之间的差 [1] 于数学模型、基于知识和基于信号处理三类 . 近年来 异, 然后针对残差信号进行故障诊断分析. 粒子滤