•  
  •  
 

Keywords

inverter; open‑circuit fault diagnosis; fast Fourier transform (FFT); improved Takagi‑Sugeno (T‑S) fuzzy neural network; adaptive Levenberg‑Marquardt (LM) learning algorithm

Abstract

Aiming at overlap and fuzziness between fault boundaries, faults, and characteristics under load disturbances and measurement noise influence when the inverter is in an open‑circuit state,, an inverter open circuit fault diagnosis model built upon the fast Fourier transform (FFT) and improved Takagi‑Sugeno (T‑S) fuzzy neural network (FNN) integration model is proposed based on the analysis of the characteristics of the inverter power tube open circuit fault. Firstly, fault characteristics are extracted when different types of open‑circuit faults occur in the power tubes according to the three‑phase output current waveforms of the inverter analyzed by the FFT. Secondly, the membership function layer of the antecedent network of the T‑S fuzzy neural network is designed by using the rule self‑splitting technology and fuzzy C‑means, and the parameters of the T‑S network are trained by leveraging the adaptive Levenberg‑Marquardt algorithm. The trained T‑S network is used to realize the diagnosis of multiple fault types and positions of the inverter power tubes. The example results show that the diagnostic accuracy of the proposed model is up to 96%, which can significantly improve the problems existing in the open‑circuit fault diagnosis of inverter power tubes.

DOI

10.19781/j.issn.1673-9140.2023.06.008

First Page

76

Last Page

86

Download

Share

COinS