Diagnosis of High-Resistance Connection Faults in PMSMs Based on GMR and Deep Learning

The high resistance connection (HRC) is a typical type of machine winding faults, which is caused by material fatigue and excessive heat in the machine windings. This fault can lead to a temperature rise or even catch a fire if left untreated. HRC fault diagnosis of permanent magnet synchronous motors are of great interest as they are widely utilized in modern industry. At present, HRC faults are mainly diagnosed by observing voltage and current signals. These methods are effective but sometimes invasive if they are not installed in the motor. This article proposes a noninvasive method, which utilizes giant magnetoresistance (GMR) sensors to collect stray magnetic field signals. The location of GMR sensors is determined by finite element Maxwell simulation, and three GMR sensors are installed in the test motor to monitor the stray magnetic field of the motor. Test data are processed by deep learning to locate and quantitatively analyze HRC faults in the motor. Experimental results show that the proposed method is effective in terms of the accuracy and fault identification. This method has potential applications for in situ electric motors.