Linear Active Disturbance Rejection Control for Propeller-Driving System based on Model Information Compensation
Baoqing Zhang, Biao Wang, Zhennan Yao, Chaoying Tang · 2020
In order to get a better performance from the Linear Extended State Observer (LESO), a Linear Active Disturbance Rejection Control (LADRC) strategy is proposed based on the Unscented Kalman Filter (UKF) identification. The UKF, with high accuracy of estimation for nonlinear distribution, is applied to identify the system information so as to provide the design parameter b0of the LADRC. In order to avoid the problem that the covariance matrix is non-positive and improve the numerical robustness, singular value decomposition is usedto instead of Cholesky decomposition in the UKF. It improves the ability of the LESO, and hence the performance of LADRC by reducing the uncertainty of the controlled system. Taking into account sensor noise, a tracking-differentiator is introduced to the LESO to get a better control performance. Both the LADRC based on the UKF identification and the traditional LADRC are applied to the propeller-driving system and compared in this paper. The simulation and experiment results both show that the proposed method has better control effect compared with the traditional LADRC, and the estimation ability of the LESO is stronger.