Motor driving system converter fault diagnosis method based on adaptive sparse filtering

Abstract

The invention discloses a motor driving system converter fault diagnosis method based on adaptive sparse filtering, and belongs to the field of driving system fault diagnosis. According to the invention, an unsupervised learning algorithm is applied to an application scene of converter fault diagnosis, effective features are automatically extracted from original data, and the problem of manual feature design based on expert knowledge is solved; meanwhile, in consideration of current period change caused by different rotating speed working conditions, rotating speed feedback is introduced, secondary sampling is carried out on current sampled at a fixed frequency, it is guaranteed that the length of a signal input into the deep sparse filtering network is a fundamental wave period, redundant information in original data is better removed, the calculation burden is relieved, and the calculation efficiency is improved. And the accuracy and rapidity of the diagnosis algorithm are improved to a certain extent.

Description

technical field [0001] The invention belongs to the field of drive system fault diagnosis, and more specifically relates to a fault diagnosis method for a converter of a motor drive system based on adaptive sparse filtering. Background technique [0002] The motor drive system is the link of energy conversion and plays an important role in many fields such as aerospace, military, electric vehicles, and ship propulsion. In the motor drive system, the converter is most prone to failure due to the harsh working environment: high dv / dt, overvoltage, thermal stress, etc., among which short circuit and open circuit faults are the most common types of faults. Once a short-circuit fault occurs, it may cause abnormal overcurrent in a very short time, such as 10 μs, which may cause direct damage to the converter. Typically, there are hardware methods in commercial products to isolate short circuit faults, such as fuses, circuit breakers or di / dt feedback methods. However, open-circu...

AI Technical Summary

Problems solved by technology

Common technologies mainly include methods based on support vector machines (SVM), methods based on k-nearest neighbors (KNN), methods based on fuzzy logic, etc., and they usually require data such as fast Fourier transform (FFT) and wavelet changes The processing method to design features still needs expert knowledge, the degree of intelligence needs to be improved, and the generalization ability is weak

Images