电机故障诊断的多传感器数据融合方法

Motor Fault Diagnosis with Multisensor Data Fusion

电机及其运行环境的复杂性决定了电机故障诊断也非常复杂 ,尽管随着检测技术、信号处理技术、智能技术的进步 ,故障诊断技术得到了很大的发展 ,但是目前的故障诊断技术仍因为各种原因存在着很大的不确定性。目前诊断技术依然是基于单个参数 ,如电流、振动、温度、润滑油成分所能携带的故障特征来进行诊断 ,但是因为模型或者环境的不确定性导致这些诊断结果模糊不清甚至错误。分析了传统故障诊断方法中存在的不确定性 ,并介绍多传感器数据融合的方法来处理由于单个参数带来的诊断的不确定性 ,同时介绍一个数据融合故障诊断系统 (fusion diagnosis system,FDS)的结构模型 ,并分析这个结构在应用中的关键问题。

AC motor fault diagnosis is a field of research with much uncertainty. Nowadays diagnosis is always based on signature of single parameter like vibration, curre nt, temperature, or chemical components, and because of the uncertainties of mod eling environment and signal processing method, decisions we make are sometimes even wrong. This paper analyzes the uncertainties in traditional fault diagnosi s method, and introduces the idea of using multisensor data fusion to handle the se uncertainties. Several parameters other than single one will be used in a fus ion system. A fusion diagnosis system (FDS) structure is introduced and critical issues in practice are analyzed.