基于粒子群优化的故障特征提取技术研究  

FAULT-CHARACTERISTIC EXTRACTING TECHNOLOGY BASED ON PARTICLE SWARM OPTIMIZATION

作　　者：潘宏侠[1] 黄晋英[1] 毛鸿伟[1] 刘振旺[1] 

机构地区：[1]中北大学机械工程与自动化学院,太原030051

出　　处：《振动与冲击》2008年第10期144-147,共4页Journal of Vibration and Shock

基　　金：国家自然基金项目(50575214);山西省自然科学基金(2007011070)

摘　　要：齿轮传动箱的故障征兆,可以通过不同的特征参量表现出来。传动箱工作过程中,由于响应信号成分复杂,提取其敏感的故障特征信息非常困难。故障程度、部位和类型等对特征参量的敏感程度差别很大,通过传统的特征提取和分析方法可以建立庞大的特征参量集。如何从众多的特征参量中确定可靠有效的故障特征参量,如何根据故障对特征参量的敏感程度优化筛选特征参量集,是实现实时在线故障诊断亟待解决的一个问题。针对传动箱故障特征选择问题,提出了基于粒子群优化(PSO)算法的特征选择方法。将粒子群优化技术用于研究传动箱振动响应信号的分析与处理,用于故障诊断特征参量集的提取与优化,形成了适合该齿轮传动箱的有效故障特征参量,从而建立了与齿轮传动箱故障现象密切相关的特征参量集。把此算法应用到齿轮传动箱故障诊断中,结果证明,该算法有很好的效果,提高了诊断精度,比常用的梯度下降算法具有更快的优化速度。The failure feature of a gearbox can be indicated using different characteristic parameters.In the work process of a gearbox,because the response signal is very complex,it is difficult to extract its sensitive fault attributive information.The sensitive level of fault,fault position and fault type are very different,so the characteristic parameter set constructed by the traditional characteristic extraction and analysis method is voluminous.Therefore,how to define the reliable and effective fault characteristic parameter set and how to optimize the parameter set using the sensitive level are the problems to be solved in order to realize real time and online fault diagnosis.Here,the characteristic extracting method base on PSO is presented for the gear box failure characteristic selection.Then,the technology is used to analyse and process the vibration response signal of a gear box,extract and optimize the fault characteristic parameter set.Finally,the parameter set osculationally related to the gear case's faults is constructed and it is used for fault diagnosis.The diagnosis results show that the PSO algorithm has good effects,higher diagnosis precision and fast optimal speed than the traditional genetic algorithm.

关 键 词：粒子群优化 故障诊断 特征提取 传动箱 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TM63[自动化与计算机技术—控制科学与工程]