Fault Detection and Diagnosis of Sensors in Induction Motor using Optimized Deep Learning Classifier

Induction motor (IM) is essential in industrial applications, and their reliable operation is critical for minimizing downtime and optimizing performance. However, IM is susceptible to various faults, including electrical and sensor-related issues, which significantly impair motor performance. This includes faults in key sensors such as speed sensors and current sensors, which leads to inaccurate measurements and affect motor control. Therefore, this paper presents an innovative approach for detecting and diagnosing sensor and electrical faults in IMs using an advanced Deep Learning (DL) approach. The proposed approach Whale Optimized Algorithm (WOA) based Bidirectional Long Short-Term Memory Network (BiLSTM) based DL classifier, employed for using hyper parameter tuning and classify the faults accurately. This study utilizes the publicly available fault diagnosis of electric drives dataset, which contains sensor signals such as current, voltage, torque, and speed under various operating conditions and fault scenarios. The motor’s raw sensor data is pre-processed to remove noise and irrelevant information, improving the quality of input data. Feature engineering is used to extract meaningful features within the processed data that highlight the distinguishing patterns associated with different fault types. The proposed system is implemented and validated using Python software, achieving an impressive accuracy of 94%, providing significant improvements in predictive maintenance and fault diagnosis for IMs.