Performance Enhancement for Active Disturbance Rejection Control
Zhuo‐Yun Nie,Yijing Ma,Qing‐Guo Wang,Ruijuan Liu
2017-01-01
Abstract:Active disturbance rejection control (ADRC) is a new kind of control technology which improves the inherent tradeoff between fast response and overshoot in the classic PID control. The basic idea of ADRC control technology is regarding the model uncertainties, external disturbances and even nonlinearity as a total disturbance, which is estimated and actively compensated by an extended state observer (ESO). After that, pole placement is easily achieved by state feedback for the desired closed-loop system. ADRC has some remarkable advantages, with small overshoot, fast respond, high precision, strong robustness, and simple tuning rules. Many ADRC applications have been reported in the literatures, such as load frequency control, magnetic rodless pneumatic cylinder, and diesel engines. Originally, ADRC is a control technique proposed by Prof. Han in the form of nonlinear feedback, including a tracking differentiator for the desired response reference and nonlinear state error feedback for the control input and a nonlinear ESO for the state and disturbance estimations. However, complex control structure and nonlinear parameter tuning make it hard to implement with digital computer and limit its practical application. To simplify the tuning process, Gao proposed the linear active disturbance rejection control (LADRC) where linear ESO and state feedback are used. Furthermore, bandwidth parameterization method is proposed to reduce the number of parameters for ADRC to two bandwidth parameters, which are closely related to the tracking and disturbance rejection performance of the controlled system. Tan shown that linear ADRC structure can be changed to a two-degree-of-freedom internal model control (IMC) structure. The analysis of LADRC can be done via the IMC framework by tuning two time constants of the setpoint filter and the disturbance rejection filter in IMC. Although many remarkable applications and improvements are made to ADRC, the existing ADRC design and parameter tuning methods still have limitations. Firstly, we know that ADRC is independence of accurate mathematic model, but it demands the accurate relative degree for the extended state observer design. When the relative degree of the plant is changing, it is necessary to redesign ESO and controller parameters. Second, there is a strict requirement on the minimum-phase (MP) plant or non-minimum phase (NMP) plant because the designs for these two cases are fundamentally different. If uncertainties cause right-half plane (RHP) zero involved, the system would become unstable. This paper first introduces an integral action in the control structure of ADRC to improve the tracking error. For the uncertainties that would cover RHP zeros, full-dimension ESO is used in ADRC, which will also allow relative-degree changing. The control system can be simply tuned by bandwidth-parametric method with better performance. Finally, the validity of the proposed method and its advantages are demonstrated through the simulations of comparative examples.