Fault Diagnosis for a Type of Electrical System with Noise Via Compressed Sensing
Ye Beilin1, Liang Kaihao2, Xiong Pingyuan3
1. Polytechnic School of Conghua, Guangzhou 510920; 2. Department of Mathematics, Zhongkai University of Agri. & Engi.,Guangzhou 510225; 3. Faculty of Mechanical & Electrical Engineering, Zhongkai University of Agri. & Engi.,Guangzhou 510225
Abstract：To tackle the problem of fault diagnosis in electrical systems with noise, a diagnosis method based on compressed sensing was proposed, which regarded fault diagnosis as a mapping from fault message set to fault types. According to such principle, a l0 norm minimization model for fault diagnosis was established, and it was transformed to l1 norm minimization. To solve the model, constructed a measurement matrix satisfying Gaussian distribution firstly, and then solved l1 norm minimization problem using IRLS algorithm to recover sparse fault signal, and finally decided fault types according to fault signal. Fault signal is exactly recovered after 3 iterations in this method; moreover, the error between setting and recovered signal is stably converged at 1%.
叶北林， 梁凯豪， 熊平原. 基于压缩感知的一类带噪声电气系统的故障诊断[J]. 电气技术, 2015, 16(8): 29-33.
Ye Beilin, Liang Kaihao, Xiong Pingyuan. Fault Diagnosis for a Type of Electrical System with Noise Via Compressed Sensing. Electrical Engineering, 2015, 16(8): 29-33.
2015, Vol. 16 
