Self-Learning Direct Flux Vector Control of Interior Permanent-Magnet Machine Drives

This paper proposes a novel self-learning control scheme for interior permanent-magnet synchronous machine (IPMSM) drives to achieve the maximum-torque-per-ampere (MTPA) operation in the constant-torque region and voltage-constraint MTPA (VCMTPA) operation in the field-weakening region. The proposed self-learning control (SLC) scheme is based on the newly reported virtual-signal-injection-aided direct flux vector control. However, other searching-based optimal control schemes in the flux-torque (f-t) reference frame are also possible. Initially, the reference flux amplitudes for MTPA operations are tracked by virtual signal injection and the data are used by the proposed SLC scheme to train the reference flux map online. After training, the proposed control scheme generates the optimal reference flux amplitude with fast dynamic response. The proposed control scheme can achieve MTPA or VCMTPA control fast and accurately without accurate prior knowledge of machine parameters and can adapt to machine parameter changes during operation. The proposed control scheme is verified by experiments under various operation conditions on a prototype 10 kW IPMSM drive.