A Novel Approach Applied to Transient Short-Circuit Diagnosis in TIMs by Piezoelectric Sensors, PCA, and Wavelet Transform

• Published in: IEEE sensors journal Vol. 23; no. 8; p. 1
• Main Authors: Lucas, Guilherme Beraldi, De Castro, Bruno Albuquerque, Ardila-Rey, Jorge Alfredo, Glowacz, Adam, Leao, Jose Vital Ferraz, Andreoli, Andre Luiz
• Format: Journal Article
• Language: English
• Published: New York IEEE 15.04.2023; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Acoustic emission; Algorithms; CCMV; Circuits; Classification; Coils (windings); Cross correlation; Discrete wavelet transforms; Emission analysis; Fault diagnosis; Indexes; Induction motors; Industrial applications; PCA; Piezoelectric sensors; Piezoelectric transducers; Principal component analysis; Principal components analysis; Sensors; Short circuits; Signal processing algorithms; Transient analysis; Wavelet Transform; Wavelet transforms; Winding
• ISSN: 1530-437X; 1558-1748
• DOI: 10.1109/JSEN.2023.3252816

Non-invasive fault diagnosis of three-phase induction motors (TIM) is widely used in industrial applications to ensure the integrity of processes. Among different types of TIM failures, transient inter-turn short circuits (ITSCs) are incipient stator winding faults characterized as short circuits between two or more turns of the coils, that can lead the winding to progressive deterioration, and consequently, the TIM to total failure. In this context, this article proposes a novel approach by using piezoelectric transducers, which performs the transient ITSC detection, phase identification, and magnitude classification by using the Acoustic Emission (AE) technique. To accomplish this analysis, a new statistical index based on the cross-correlation function was proposed to detect the ITSC and classify its magnitude. Besides, Wavelet Transform and Principal Component Analysis (PCA) stood out as promising tools to identify which phase was affected by the short circuits. A TIM was subjected to ITSCs and the experimental results showed that the proposed algorithm successfully performed the transient ITSC detection, phase identification, and evolution classification. Additionally, this work improve the capabilities of traditional AE systems, since no AE signal processing algorithm has ever been proposed for a comprehensive diagnosis of transient ITSC.