Deep Transfer Learning-Based Broken Rotor Fault Diagnosis For Induction Motors

Due to their starting and running torque needs as well four-quadrant operation, modern industrial drives utilise induction motors (IM). Failures in the rotor bars of motor can be found using voltages currents each three phases acceleration velocity signals. For diagnosis quantity broken for a failed IM, conventional signal processing-based feature extraction techniques machine learning algorithms have been applied past. The number is determined this study by looking into novel technique. aforementioned aims, specifically, deep methodologies are studied. In order do this, convolutional neural network (CNN) transfer described. Initially, bandpass filter used denoising, then signals transformed continuous wavelet transform create time-frequency pictures (CWT). collected images classification support vector (SVM) classifier, fine-tuning pre-trained ResNet18 model. Metrics performance evaluation employ categorization accuracy. Additionally, results demonstrate that features recovered from mechanical vibration current yield greatest accuracy score 100%. Nonetheless, comparison with publicly available also done. comparisons proposed strategy outperforms compared methods terms scores.