摘要
在通常的故障诊断中,一个约定的假设是系统在同一时刻至多只有一个故障发生。然而,对于拥有冗余设计或在工作过程中几乎没有维修和保养机会的系统(如大型航空航天器等),该假设往往并不成立。另外,对于系统的关键性设备,获得足够的典型故障样本通常非常困难。针对这种情况,本文以模拟电路系统为研究对象,提出了一种多故障诊断方法。首先建立了系统的多故障模型,在此基础上用小波变换作为预处理器进行特征提取,最后采用多类支持向量机(SVM)实现故障识别。仿真结果验证了该方法的可行性。
A common simplified assumption made by the existing fault diagnosis methods is that there exists, at most, a single fault in the system at any given time. However, this assumption does not hold true for complex systems with redundancy design and/or systems with little or no opportunity for maintenance during operation. Furthermore, it is difficult to obtain enough and complete fault samples for the key equipment of a system. Aiming at these circumstances, we present a multiple fault diagnosis approach for analog circuit. Firstly, we build the multiple fault model of the system, and based on this we extract fault feature using wavelet transform as preprocessor. The fault diagnosis is implemented by multi-class SVM finally. Simulation experiments verify the effectiveness of the proposed method.
出处
《仪器仪表学报》
EI
CAS
CSCD
北大核心
2007年第6期1029-1034,共6页
Chinese Journal of Scientific Instrument
基金
国防基础科研项目(A1420061264)资助
关键词
多故障诊断
小波变换
多类支持向量机
multiple fault diagnosis
wavelet transform
multi-class support vector machine
