Research on fault diagnosis method of asynchronous motor based on Park-WPT and WOA-LSSVM

Gong-kuang zidonghua

RONG Xiang2,3¹  HUI Ali1¹  CHEN Wenya2,3²  WEI Lipeng2,3³  LU Weiqiang1³ 

[1] 1.College of Electrical and Control Engineering, Xi'an University of Science and Technology, Xi'an 710054,China;2.CCTEG Changzhou Research Institute, Changzhou 213015, China;;

关键词: asynchronous motor;    multiple fault diagnosis;    park vector transform;    wavelet packet transform;    whale optimization algorithm;    least squares support vector machine;   

DOI  :  10.13272/j.issn.1671-251x.2021070035

来源: DOAJ

【 摘 要 】

In order to solve the problems of poor precision and high cost of the existing motor multiple fault diagnosis technology, the rotor broken, air gap eccentricity and their mixed faults of asynchronous motor are studied based on three-phase stator current signals, and a fault diagnosis method of asynchronous motor based on Park-WPT (Park-wavelet packet transform) and WOA-LSVM (whale optimized algorithm-least squares support vector machine) is proposed. The collected three-phase current signals are preprocessed through Park vector transformation, the signal characteristics are extracted according to the distortion rate of the elliptical trajectory and the signal characteristics are taken as the first type characteristic quantity. The wavelet packet transformation is performed on the Park vector modulus square spectrum so as to obtain the energy value of its decomposition coefficient as the second type characteristic quantity. The mechanism of WOA's shrinkage surrounding prey and spiral updating prey position is used to optimize the regularization parameters and kernel width in LSSVM, and a fault diagnosis model based on WOA-LSSVM is established based on the extracted two types of characteristic signals. The experimental results show that the single characteristic extraction algorithm based on Park vector transform or wavelet packet transform has poor recognition effect on mixed faults, and the recognition rates of fault characteristics are 73.75% and 88.33% respectively. The recognition rate is improved to 97.08% by combining the two types of characteristics. WOA-LSSVM has a faster optimization speed and a higher fault diagnosis accuracy rate. Its overall performance is better than PSO (particle swarm optimization) algorithm, GWO (grey wolf optimization) algorithm and GA (genetic algorithm) optimized LSSVM.

【 授权许可】

Unknown