Publication

Abstract : Accurate classification of faults in induction motors is crucial for minimizing downtime, enhancing energy efficiency and lifespan, optimizing repair work, and ensuring safe operation. Accurate classification can be achieved using infrared images. Machine learning models, such as Convolutional Neural Networks (CNNs), have demonstrated superior performance in image classification problems. However, classification accuracy is highly dependent on the selection of activation functions, hyperparameter optimization techniques, and the quality of images. This necessitates selecting the best combination of activation functions and hyperparameter optimization techniques. Further, the image quality can be significantly improved by image processing techniques like image segmentation and enhancement. Therefore, this paper proposes a CNN-based fault classification model for induction motors. The proposed model selects the best combination of activation functions and the hyperparameter optimization technique through performance analysis. Further, the proposed model uses the Simple Linear Iterative Clustering (SLIC) algorithm for image segmentation. It divides the fault images into superpixels, effectively isolating key regions that contain important fault features. Additionally, the Contrast Limited Adaptive Histogram Equalization (CLAHE) technique is applied to enhance the image contrast, making subtle fault patterns more discernible. The results show that the proposed model improves fault classification accuracy to 100%.