| Title |
A Study on the Practical Operation of DSP for Motor Control Embedded with Fault Diagnosis and Predictive Algorithm |
| Authors |
송승민(Seung-Min Song) ; 한지훈(Ji-Hoon Han) ; 최의진(Eui-Jin Choi) ; 박종훈(Jong-Hoon Park) ; 홍선기(Sun-Ki Hong) |
| DOI |
https://doi.org/10.5370/KIEE.2024.73.1.97 |
| Keywords |
Motor fault diagnosis; Motor controller; Deep learning; Mechanical fault; Auto-encoder; DSP |
| Abstract |
In the fault diagnosis in electric motors, additional device is commonly required. This study proposes a solution to diagnose mechanical faults in electric motors by integrating a deep learning-based algorithm into the motor controller without additional hardware. Operating within Microcontroller Unit (MCU) memory constraints, the algorithm ensures comparable classification performance to existing methods using current signals. The chosen approach employs an auto-encoder in the MCU, adjusting neuron counts based on current signal sampling frequency and optimizing the model structure for a balance between memory usage and diagnostic success rate. The proposed model's architecture prevents interference with the control routine, showcasing stable control performance and effective fault diagnosis within the integrated system. |