基于压缩感知的异步电动机故障诊断数据压缩与重构  

Compression and Reconstruction for Fault Diagnosis Data of Asynchronous Motor Using Compressed Sensing

作　　者：刘建林[1] LIU Jian-lin(Department of Electrical Engineering,Hunan Mechanical&Electrical Polytechnic,Changsha 410151,China)

机构地区：[1]湖南机电职业技术学院电气工程学院

出　　处：《湖南师范大学自然科学学报》2018年第5期88-94,共7页Journal of Natural Science of Hunan Normal University

基　　金：国家自然科学基金资助项目(51407064);湖南省自然科学基金资助项目(2018JJ5030)

摘　　要：为提高异步电动机故障诊断过程中所采集状态信息的传输效率及诊断结果的可靠性,提出了一种采用奇异值分解的压缩感知优化方法.该方法首先对测量矩阵进行奇异值分解,然后将优化后的测量矩阵和测量向量用于压缩感知贪婪类迭代算法中,再通过精确的夹角余弦法筛选出与残差最为匹配的候选集原子,并将此原子用于稀疏信号的重构,最后得到估计的故障诊断数据信号.仿真结果表明:在相同的信号稀疏度或测量数目下,提出的优化算法相比传统的压缩感知算法能极大提高远程故障诊断数据信号的重构精度,这对实际工程中后期故障的有效去除具有重要意义.In order to improve the transmission efficiency and reliability of the diagnosis results in asynchronous motor fault diagnosis process,a compression sensing optimization method using singular value decomposition was proposed in this paper.The method first decomposed the singular value of the measurement matrix,and then used the optimized measurement matrix and measurement vector to greedy iterative algorithm of compressed sensing.Next,the most suitable candidate atoms are selected by the accurate method of angle cosine,and the atom was used for the reconstruction of the pulse signal.Finally,the estimated fault diagnosis data signal was obtained.The simulation results show that the proposed optimization algorithm can greatly improve the reconstruction accuracy of remote fault diagnosis data signals compared with the traditional compression sensing algorithm under the same signal sparsity or number of measurement,which is of great significance for the effective removal of faults in the later stage of practical projects.

关 键 词：压缩感知 异步电动机 奇异值分解 观测矩阵 

分 类 号：TP391.5[自动化与计算机技术—计算机应用技术]