2023 Volume 12 Issue 4 Pages 816-825
Position-sensorless positioning servo systems for interior permanent magnet synchronous motors (IPMSMs) have been developed to achieve dimension and cost reduction. In these systems, parameter mismatch between the IPMSM, position controller, and position estimator due to thermal variation and aged deterioration is inevitable. To solve this problem, a parameter identification method based on an adaptive scheme has been proposed. However, to use the adaptive scheme, this method can only be applied under no-load conditions, and it is difficult to compensate for parameter variations during actual operation, i.e., under load conditions.
This paper proposes a novel learning-based position control in position-sensorless positioning servo systems. In the proposed method, a feedfoward controller established via learning adaptively compensates the parameter fluctuations in these systems. As learning progresses, the transient response of position control is improved while ensuring robustness to disturbance torque. The effectiveness of the proposed position control system is demonstrated via experiments.