关键词: 永磁同步电机, 数字孪生, 高保真建模, 多新息理论, 参数辨识

Abstract:

Conventional mechanism models often suffer from model mismatch due to the absence of nonlinear mechanical damping representation and the dynamic drift of electromagnetic parameters. To address this issue, a high-fidelity dynamic digital twin modeling method for permanent magnet synchronous motors is proposed. In the static mechanism modeling dimension, a friction energy decoupling strategy is proposed, which fundamentally overcomes the inherent conflict between steady-state accuracy and transient response from a physical logic perspective. For dynamic model updating, a distributed cascaded sensing architecture based on the multi-innovation recursive least squares and extended Kalman filter (MIRLS-MIEKF) algorithms is proposed. This architecture enables the dynamic observation of time-varying motor parameters and the real-time correction of the digital twin model. Simulation results indicate that the proposed hybrid mechanism and data-driven digital twin system achieves high-precision tracking of actual dynamic trajectories under various nonlinear operating conditions, thereby improving physical fidelity. Compared with conventional mechanism models, the mean absolute error and root mean square error of the speed trajectory are reduced by up to approximately 66% and 63.2%, respectively, providing a high-fidelity modeling foundation for the monitoring and control of PMSM.

Key words: permanent magnet synchronous motor, digital twin, high-fidelity modeling, multi-innovation theory, parameter identification