کنفرانس بین المللی ماشین و محرکه های الکتریکی

صفحه اصلی / The 5th International Conference on Electrical Machines and Drives

Deep CNN-Based Method for Detecting Mechanical and Rotor Bar Faults in Induction Motors

نویسندگان :

Seyedjavad Mirmohammad Meiguni¹ Abolfazl Vahedi² Akbar Rahide³

1- دانشگاه علم و صنعت 2- دانشگاه علم و صنعت 3- دانشگاه صنعتی شیراز

کلمات کلیدی :

Induction Motor،Condition monitoring،Fault Detection،Mechanical Fault،convolutional Neural Network

چکیده :

Given the widespread use of induction motors and their critical role in industry, condition monitoring and fault detection in these motors is of utmost importance. Among all faults associated with induction motors, mechanical faults account for the largest share. So far, various methods have been proposed and implemented for detecting each or several of mechanical faults in induction motors. However, a major limitation of these methods is that they are often not automatic and rely on personal experience. Additionally, in automated and artificial intelligence-based methods, not all faults are addressed, and faults are typically identified on a case-by-case basis. In this paper, a comperehensive and intelligent fault detection method based on Deep 2-D Convolutional Neural Networks (DCNN) is proposed, which has several unique features: 1. Unlike previously proposed methods, it is capable of detecting mechanical faults as well as broken rotor bar faults in squirrel-cage motors. 2. This method only uses the Fast Fourier transform of the three-phase current signal 3. The method achieves remarkable accuracy in detecting and classifying the aforementioned faults. The proposed method has been evaluated on a laboratory test bench consisting of a 1.5 kW three phase induction motor under various loading conditions. The obtained results demonstrate the effectiveness of the method in detecting the aforementioned faults