Different Induction Motor Faults by New Proposed Random Forest Method

Abstract

Induction motors (IM) are widely used in industry. Failures in asynchronous motors cause disruptions and interruptions in production processes. Due to this situation, economic losses are experienced. Monitoring the induction motor status and monitoring the symptoms before the failure occurs is a matter of great importance in the industry. In this study, 8 different situations that may occur in the motor were monitored through the acceleration and sound data obtained from the induction motor. The feature vector was created with the Short-Term Fourier Transform (STFT) method on the acceleration and sound data obtained from the engine. The feature vectors were classified using the Random Forest (RF) method. The feature vectors created from the acceleration and sound data were also analyzed separately and the classification performance was examined. In addition, a new RF algorithm based on weight values using the Gini algorithm has been proposed. With this algorithm, the traditional RF algorithm has been developed and the success rates have been increased. In classical RF classification based on acceleration and sound data, 89.9% accuracy was achieved. The success rate of the proposed model was 95.7%. This shows that the proposed model successfully detects all types of faults in asynchronous motors. In addition, when we compared in terms of time, it was observed that the proposed model produced faster and more accurate results both in fault detection and in the production maintenance phase.