DATA FUSION BASED MULTIMODAL FAULT DIAGNOSIS IN PERMANENT MAGNET SYNCHRONOUS MOTORS

In this study, a deep learning-based method is proposed for the classification of inter-turn and inter-coil short circuit faults occurring in three-phase Permanent Magnet Synchronous Motors (PMSM). Three-phase current and vibration signals are used by multi-mode data fusion and a Convolutional Neural Networks (CNN) model. The spectrograms used as input in the CNN model are obtained using Short Time Fourier Transform (STFT). Using the proposed method, faults are classified with high accuracies (inter-turn faults with 100% accuracy, inter-coil faults with 98.95% accuracy). The obtained results show that the proposed multi-mode fusion approach provides high success in both fault detection and fault severity classification.