Active disturbance rejection control: applications, stability analysis, and tuning method

Abstract

Active disturbance rejection control (ADRC), an emerging control technique, not only has a simple control structure but also has the advantage of being robust against plant uncertainties and external disturbances. In this thesis, the application of the ADRC is extended to two new areas: automated steering control for lane keeping in autonomous vehicles, and position control of hydraulic actuators. The thesis addresses two challenges in the above applications. The first is the stability analysis of the vehicle system controlled by the ADRC. Because the vehicle model is nonlinear and complex, using Lyapunov’s second method to study the stability is challenging. The second challenge is the tuning of the ADRC for controlling hydraulic actuators. It is found that the ADRC tuned by the widely used bandwidth tuning method cannot provide the desired tracking performance of hydraulic actuators. To solve these challenges, the concept of Lyapunov exponents is used to investigate the stability of the ADRC system, and the quantitative feedback theory (QFT) is adopted in a novel manner to tune the ADRC for the position control of hydraulic actuators. With respect to the automated steering controller design, the models describing the vehicle dynamics are first developed. Second, the ADRC steering controller is designed and the stability of the ADRC system is investigated using the concept of Lyapunov exponents. Third, simulations and experiments are conducted to validate the effectiveness of the ADRC controller. It is shown that the vehicle controlled by the ADRC controller performs the lane keeping successfully. The ADRC controller is simple, without requiring complex calculations and real-time measurements, and it is easy to implement in practice. On the position controller design for the two hydraulic actuators, including the electro-hydrostatic actuator and the electro-hydraulic actuator, the ADRC is first transformed from the state space to the frequency domain. Then, the decomposed controller and prefilter are tuned in the framework of the QFT to meet the prescribed design specifications for reference tracking and stability. Both the simulation and experimental results demonstrate that the actuators track the desired positions closely. The proposed tuning method for the ADRC provides an easy-to-use and effective tool for controller design in practical applications.