A novel key performance analysis method for permanent magnet coupler using physics-informed neural networks

The non-contact transmission product permanent magnet coupler (PMC) has been widely used in industry due to its advantages such as low noise and vibration, high efficiency, high reliability, and overload protection. Owing to its complex electromagnetic behaviors, the accurate computation of key performances such as magnetic vector potential and torque is essential for optimizing its transmission efficiency. However, traditional calculation methods are based on analytical or simulation models, either imprecise or computationally intensive, hindering subsequent optimization design modeling. To address the above issue, this paper proposes a novel method based on physical-informed neural networks (PINN) to calculate the PMC performances with high accuracy and low computational cost. PINN integrates prior knowledge into the deep neural network’s loss functions to establish the model and accurately predict the PMC’s performance parameters. Experimental results demonstrate that PINN outperforms traditional calculation methods regarding feasibility, validity, and accuracy. Overall, PINN combines data-driven models with prior knowledge to achieve a data–knowledge dual driven, providing a new approach for optimizing PMC structure design.