基于混沌和数据挖掘的电子系统故障诊断  

Chaos and Data Mining Based Fault Diagnosis for Electronic System

作　　者：马红光[1] 韩崇昭[1] 王国华[2] 许剑锋[2] 朱小菲[2] 

机构地区：[1]西安交通大学电信学院,西安710049 [2]第二炮兵工程学院一系,西安710025

出　　处：《数据采集与处理》2004年第3期273-277,共5页Journal of Data Acquisition and Processing

摘　　要：提出了一种基于相空间重构和数据挖掘算法的复杂电子系统故障诊断方法 ,混沌时间序列的相空间重构采用了复自相关和Γ -test的嵌入维、时间延迟联合算法 ,所得到的结果为准最佳的嵌入维、时间延迟。采用计算相空间中各个单位质点绕其质量中心的旋转半径的数据挖掘算法估计被测对象的工作状态 ,用统计学中的 3 σ准则识别被测系统的故障参量。实验结果表明 ,该故障诊断方法可准确地检测出复杂电子系统中的故障现象。A fault diagnosis method based on the phase space reconstruction and data mining approaches for the complex electronic system is presented. The approach for the phase space reconstruction of chaotic time series is a combination algorithm for multiple autocorrelation and Γ-test, and the quasi-optimal embedding dimension and time delay can be obtained by using the algorithm. The data mining algorithm calculates the radius of gyration of unit-mass point around the centre of mass in the phase space, and the operating status of the tested system is estimated by using the algorithm. The 3σ criterion in statistics distinguishes the fault parameter from the chaotic time series output by the tested system. Experimental results show that the fault diagnosis method can correctly detect the fault phenomena of the complex electronic system.

关 键 词：数据挖掘 故障诊断 电子系统 混沌时间序列 相空间重构 

分 类 号：TP206.3[自动化与计算机技术—检测技术与自动化装置]