Fault Diagnosis System of Hall Sensor in Brushless DC Motor based on Neural Networks Approach

Abstract

Hall sensors are commonly used or built in a motor system. The functionality of the hall sensor is to detect the speed and position of the motor. The normal operation of motors is affected by hall sensor’s fault. A fault diagnosis system is implemented in the motor system is commonly used to detect faults. In the industry, traditional methods such as the state-sensitive method or edge-sensitive method are widely implemented. Traditional methods have limitations such as complexity for implementation in other models and less robust. This paper proposed a fault diagnosis system based on the neural network approach. The characteristics of different types of neural networks were studied. Different types of neural networks were implemented, not every neural network variant was able to achieve a decent performance for the fault diagnosis system. The results were shown that the fault diagnosis system based on both CNN and DNN effectively determine faults and achieve accuracy above 95%.