Inductance and Core Loss Modeling and Estimation of Permanent Magnet Synchronous Motor with Subspace Learning-Based Flux Linkage Denoising

In the field of control and optimization for permanent magnet synchronous motors (PMSM), it is challenging to accurately estimate the inductances along the motor’s d-axis and q-axis, as well as to quantify the complex effects of core loss. This study proposes a inductance and core loss estimation method for PMSM based on flux linkage denoising and neural network modeling. Specifically, an equivalent circuit model of the PMSM considering the core loss effect is established. A noise removal technique based on subspace learning is introduced to denoise the flux linkages data. Furthermore, the flux linkage equations considering the core loss effect are integrated into the neural network to estimate the motor’s inductances and parameters related to core loss. Experimental results demonstrate that the proposed neural network can accurately estimate the inductances and core loss-related parameters of PMSM, successfully addressing the issue of quantifying core loss effects, which validate effectiveness and robustness of the proposed approach.