A simple signal extraction‐based online real‐time diagnosis approach for interturn short‐circuit fault of permanent magnet motor

Abstract

Summary A fast and reliable interturn short circuit (ITSC) diagnosis system is necessary to secure safe operations of permanent magnet synchronous motor (PMSM) for its widespread applications in industrial automation and electric vehicles. This paper develops a simple online real‐time technique based on signal extraction to detect the ITSC fault in the PMSM. Some popular diagnosis methods involving machine learning or instantaneous power information may require large computing resources, data storage, or extra sensing devices for data collection. This makes it difficult to fulfill online and self‐detection of ITSC faults in motor drive systems. This paper makes use of the second‐order harmonic components in dq‐axis current for diagnosis when they occur due to ITSC faults. Through the developed amplitude‐tracking algorithm, the diagnosis can be implemented by using a common digital signal processor without complex algorithms and additional hardware or sensors. Therefore, it is suitable for low‐cost and remote applications where self‐detection is needed. The proposed method is validated by finite element analysis and measurements with a hardware‐in‐the‐loop toolkit. It is found that the method successfully detects a short circuit fault with only one out of 54 turns in one phase of the motor (1.9%).

Single open‐phase fault detection and tolerant control for wye‐connected three‐phase induction machine drives

“…In general, fault detection (FD) is needed for the operation of electrical machine drives from the normal condition to the SOPF condition. Over the last decades, several model-based 26,27 and signal-based 28,29 FD approaches have been proposed by researchers. It is notable that in model-based FD methods, the difference between the estimated and actual signals is considered the FD index, while in signal-based FD techniques, the output voltages/currents of the inverter are measured, and the FD index is achieved by a signal processing strategy.…”

Magnetic tile fault detection of high‐voltage electric machines: A multiview feature selection method via consistent soft label

“…For the signal detection method, Hsu et al extract second harmonic signal components for interturn short circuit fault diagnosis. 4 For image detection method, Younus et al use a two-dimensional discrete wavelet transform combined with a feature selection method to process infrared images for mechanical fault diagnosis. 5 In high voltage motor magnet fault diagnosis, the magnet element is located in the motor core, and its signal is often submerged by environmental noise or other interference signal components.…”

“…Standard fault diagnosis techniques can be divided into signal and image detection. For the signal detection method, Hsu et al extract second harmonic signal components for interturn short circuit fault diagnosis 4 . For image detection method, Younus et al use a two‐dimensional discrete wavelet transform combined with a feature selection method to process infrared images for mechanical fault diagnosis 5 .…”

Magenetic Tile Fault Detecion of High Voltage Electitric Machine: A consistent soft-label-based Multi-view feature selection Method

“…Sometimes it is difficult to use deep learning methods for detection. The collected data will be of high dimensionality 4,5 . Therefore, it is necessary to use dimensionality reduction techniques to process data in fault detection.…”