This paper proposes an artificial intelligence (AI)-assisted maximum-torque-per-ampere control of an interior permanent magnet synchronous machine (IPMSM). For torque control of an IPMSM, the optimum current trajectories are calculated. The current components are determined such that the magnitude of the current vector is minimized for a reference torque. Especially in the field weakening range, this results in nonlinear equations that have to be solved, which is why the current components are often calculated offline and stored discretely in look-up tables (LUTs). Due to the discrete storage of data, interpolation is performed for data not included in the LUT, resulting in interpolation errors. In this work, these LUTs are replaced by neural networks, minimizing this interpolation error. In addition to a significantly improved interpolation accuracy, the required memory space on a microcontroller is also minimized.
AI-Assisted Torque Control of an Interior Permanent Magnet Synchronous Machine
2023-11-28
1489189 byte
Conference paper
Electronic Resource
English