Spatiotemporal Fusion Model Based on Gramian Angular Field for Motor Fault Identification

With the widespread application of three-phase synchronous motors in industry, the detection of electrical faults has become particularly important. To address the limitations of current fault identification methods in terms of accuracy and robustness, we propose a spatiotemporal fusion model based on Gramian Angular Field for fault diagnosis of three-phase synchronous motors. This method employs the Gramian Angular Field to convert the original electrical signal into a twodimensional image, and uses the image and the original time series signal as the spatial and temporal information input of the fault identification model respectively. Finally, a cross-attention mechanism is designed to achieve deep fusion of spatiotemporal information. Extensive experimental results on public datasets demonstrate that the proposed method has excellent fault identification accuracy and stability.