A High-Fidelity and Computationally Efficient Model for an Electrically Excited Synchronous Generator Based on Current–Flux Linkage Neural Networks

Driven by the trend of multielectric aircraft, the digitization of aircraft power systems is crucial for their effective condition monitoring and fault diagnosis. This study proposes a modelling technique for the main generator (electrically excited synchronous generator, EESG) of a three-stage aircraft synchronous generator. At present, finite element (FE) motor simulation exhibits high accuracy in the field of nonlinear simulation but is computationally intensive. Interestingly, the modelling method for the FE flux linkage–current lookup table can overcome the computational cost limitation. However, this method can only simulate accurately in the range of FE scanning conditions and has data fitting problems. Consequently, the proposed model uses FE flux linkage–current data to establish the EESG dq model and replaces the flux linkage–current lookup table with the flux linkage–current neural network. This approach enables the EESG model to perform the nonlinear simulation and prediction of unknown operating points. Finally, the accuracy, computational efficiency, and predictive ability of the proposed model are verified by FE simulation and experiments. Through testing, the proposed model can consider the nonlinear characteristics of the motor during the simulation process and has certain predictive abilities, greatly improving the simulation efficiency.