【作者】 钱朋朋； 刘金国； 张伟； 魏英姿；

【Author】 Qian Pengpeng~(1,2) Liu Jinguo~(2,1) Zhang Wei~2 Wei Yingzi~1 (1.School of Information Science & Engineering,Shenyang Ligong University,Shenyang 110159,China; 2.State Key Laboratory of Robotics,Shenyang Institute of Automation,Chinese Academy of Sciences,Shenyang 110016,China.)

【机构】 沈阳理工大学信息科学与工程学院； 中国科学院沈阳自动化研究所机器人学国家重点实验室；

【摘要】 主元分析(PCA)被广泛应用于多传感器系统的故障诊断当中。然而,PCA只适合于分析故障或干扰仅存在于某一固定尺度的情况。实际当中,工程数据往往是多尺度的。因此,本文提出了将多尺度主元分析(MSPCA)法应用于传感器的故障诊断当中。MSPCA综合了PCA去除变量间关联、小波分析提取变量特征的优势,能检测出不同尺度/频率下的故障。仿真实例表明:与PCA方法相比,MSPCA方法在进行故障检测时效果更好。更多还原

【Abstract】 Principal component analysis(PCA) is widely used in fault diagnosis of multi-sensor system.However,it is only suitable for analyzing the fault existing in a fixed scale or some specific frequency band.In practice,the engineering data are often multi-scale.Therefore,this paper has proposed the multi-scale principal component analysis(MSPCA) for sensor fault diagnosis. PCA can capture the linear correlation between variables,and wavelet analysis can analyze the signal in different scales.MSPCA composites the advantages of PCA and wavelet analysis,and it can detect the faults in different scales/frequencies.Simulation examples show that MSPCA is more effective than PCA in sensor fault detection.更多还原