Magnetic Field Analysis of Reactor Based on TL-PIDL and Time Domain PIDL

With the widespread use of deep learning in various fields, physics informed deep learning (PIDL) is proposed to be applied in a magnetic field problem. To meet the needs of practical engineering, we predict the electromagnetic response of the reactor by combining the theory related to magnetic fields with PIDL, and its feasibility is verified by comparing its predictions to those of finite element methods (FEM). Subsequently, by using transfer learning (TL), it is possible to perform quick analysis based on analytical experience before changing the geometry. Lastly, the performance of the proposed PIDL was validated using results obtained from both measurements and FEM, thus proving that the method can significantly reduce the training time while maintaining accuracy which can meet the precise and real-time needs of engineering applications.