Abstract
This paper proposes a torque ripple suppression method for dealing with the system identification model error through learning and correction. A torque ripple generated by a motor or any other source causes many problems such as noise and vibration. We proposed the periodic disturbance observer to suppress a torque ripple. This method is a relatively simple control scheme built by using a system identification model. It has a high suppression effect regardless of the order of the target frequency. However, the model error leads to an increase in the convergence time to suppression and unstable control. Therefore, there is a need to improve the robustness against modeling errors.
The proposed method corrects for the existing system model through dynamic system identification using a torque ripple component detection value and estimate value. By adding the error correction function to the periodic disturbance observer, the problem caused by the model error is solved and the robustness is improved. This paper explains the proposed method and demonstrates the effectiveness of the proposed method through simulation and experimental results.
