A Robust Deadzone Compensation Method Against Parameter Variations based on Kalman Filter and Neural Networks

Due to the skin effect, parameters of motor windings change over frequency. In order to solve the problem in traditional Kalman-filter based deadzone compensator that the performance is sensitive to parameter mismatch, a novel deadzone compensation scheme with high robustness is proposed. Firstly, a deadzone compensation scheme based on Kalman filter technology is presented, and motor parameter variation over frequency changes is studied. Secondly, the impact of motor parameter mismatch is analyzed, and the verification of the performance deterioration is done via computer simulation. Thirdly, an Adaline neural-network(NN) based robustness enhancement algorithm is proposed to analyze the current error components. The performance deterioration is compensated by using the analyzed results of the robustness enhancement algorithm. Finally, the robustness against parameter sensitivity of the proposed method under various parameter mismatch conditions is fully studied and verified.