Dynamic data reconciliation to enhance the performance of model free adaptive control

Abstract

As a novel data-driven control method, model-free adaptive control (MFAC) has a high requirement for the accuracy of the feedback signal. However, the sensor used to obtain the output inevitably contains measurement noise due to its own error or external interference, which may lead to an adverse effect on the control performance. The dynamic data reconciliation (DDR) is proposed and combined with MFAC to improve the control performance in this paper, which uses predicted output and measured data to suppress measurement noise considering Gaussian and non-Gaussian distributed measurement noise. The effectiveness of the DDR combined with MFAC (DDR-MFAC) is illustrated in the single-input single-output and multiple-input multiple-output systems with Gaussian and non-Gaussian distributed measurement noise. DDR-MFAC is also successfully applied to DC–AC converter, which improves its conversion precision.