Online Attribute Matching Based Few-Sample Data-Driven Diagnosis of Electrical Faults in PMSM Drive

In this article, an online attribute matching based few-sample data-driven diagnosis method for electrical faults in permanent magnet synchronous motor drive is proposed to improve the diagnosis precision with fewer training data and lower computational complexity. By incorporating the motor model k...

Full description

Saved in:
Bibliographic Details
Published in:IEEE transactions on power electronics 2024-02, Vol.39 (2), p.2620-2631
Main Authors: Jin, Luhan, Wang, Xueqing, Mao, Yao, Lu, Linlin, Wang, Zheng
Format: Article
Language:English
Subjects:
Attribute matching
Computational modeling
Computing time
data-driven diagnosis
Electrical faults
Fault detection
Fault diagnosis
Faults
Feature extraction
integrated attribute learner
Lookup tables
Matching
Mathematical models
Motor drives
permanent magnet synchronous motor (PMSM) drive
Permanent magnets
Solution space
Switches
Synchronous motors
Torque
Training
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this article, an online attribute matching based few-sample data-driven diagnosis method for electrical faults in permanent magnet synchronous motor drive is proposed to improve the diagnosis precision with fewer training data and lower computational complexity. By incorporating the motor model knowledge in the feature extraction process, effective and robust input features can be extracted without consuming large computing resources. Instead of directly predicting the fault mode in the original huge solution space, an integrated attribute learner is specifically designed to predict the fault attributes in the simplified solution space to reduce both network size and required training data. Finally, based on the look-up table for the fault attribute vector, an attribute matching strategy is proposed to determine the specific fault mode with the consideration of unknown faults. Comprehensive experiments verify that the proposed method can identify each fault mode efficiently with an average diagnostic accuracy of 99.12% and the computational time of 0.1 ms, which outperforms the existing data-driven diagnosis methods for motor drives.
ISSN:0885-8993
1941-0107
DOI:10.1109/TPEL.2023.3335268