Sensors fault diagnosis of bearingless permanent magnet synchronous motor based on GAF-pCNN-BiLSTM

Abstract To enhance sensor fault detection in bearingless permanent magnet synchronous motors, this paper proposes a multi-signal fusion fault diagnosis method that integrates the Gramian angular field (GAF) transformation, a parallel convolutional neural network (pCNN), and a bidirectional long short-term memory (BiLSTM) network. First, sensor data are transformed into two-dimensional (2D) images using the GAF method. These images are then processed by the pCNN to extract fault feature vectors. Finally, the BiLSTM network captures transient temporal features embedded in the fault characteristics, enabling accurate fault classification. Experimental results demonstrate that the proposed method not only ensures precise fault identification but also significantly outperforms other machine-learning-based fault diagnosis approaches in terms of classification performance, thereby confirming its feasibility and reliability.