Observer-based Disturbance Rejection Control Methodology And Performance Evaluation

In the applications of control system design, system uncertainties such as parameters perturbation, unmodeled dynamics, coupling between system states, and external disturbance, have great influence on control accuracy and system robustness, or even make the system unstable. Thus, the problem of high performance disturbance rejection control has been an extensive topic in control theory and its applications. The observer-based control methodology is an effective way in disturbance rejections. The observer is employed as the inner-loop of the control system. The input and output information of the control object is utilized to estimate the system uncertainty online,and then a disturbance compensator is introduced to increase the control accuracy.For the existing observers, disturbance observer(DOB) and extended state observer(ESO) have been widely explored. Although both of them can deal with system uncertainties, the analysis and design are conducted based on different theoretical systems. Thus, it is important to analyze their relations, differences and performances.In this thesis, researches on DOB and ESO are carried out for three kinds of systems.From the above researches, the disturbance rejection structure and control performance comparison are explored, based on which selecting guidance of observers are proposed according to the type of controlled system and objective. The main research works and contributions of this thesis are summarized as follows:(1) For non-minimum phase system with both right-half plane(RHP) zeros and time-delay, a DOB design and parameters optimization strategy is proposed systematically is this thesis. The proposed strategy synthesizes the internal stability, relative order, disturbance rejection performance, mixed sensitivity design requirements together to establish an optimization function of Q filter. The optimization problem is transformed into a standard H∞one. The solution of Q filter is optimized to guarantee the internal stability constraints by introducing an all-pass portion into virtually controlled object of H∞problem.(2) For nonlinear system with observable states, a robust DOB based disturbance rejection controller is proposed and parameters optimization strategy is investigated.First, a DOB with partly feedback linearization and a low-pass filter is proposed for nonlinear dynamic model, and an outer-loop backstepping controller is designed. By analyzing the outer-loop controller’s influence on the inner-loop observer parameters,the constraint is proposed to guarantee the robust stability of the closed-loop system.The standard H∞control method can be used to optimize the parameters of DOB.Experiment of attitude tracking on an aircraft is carried out to show the effectiveness of the proposed control strategy.(3) For nonlinear system with unobservable states, a generalized ESO design methodology is proposed in this thesis. The disturbance rejection performance of traditional ESO is analyzed to show the essence of the reconstruction strategy. The modal of equivalent disturbance is analyzed and reconstructed, based on which prior information of the equivalent disturbance can be utilized as far as possible to decrease the system uncertainty. The generalized ESO based on reconstructed system can estimate the disturbance with known modal thoroughly. Stability of the proposed ESO is analyzed along with the outer-loop feedback controller based on Lyapunov theory.(4) Since the performance of the closed-loop system depends on the disturbance estimation effect, a concept of disturbance rejection structure is proposed. Both of DOB and ESO are analyzed under this framework, based on which the observer design problem can be transformed into the structure design and parameter optimization of equivalent low-pass filter under the disturbance rejection structure. Control performance of these two methods is analyzed based on different types of controlled objects and tasks. The differences of these two methods are summarized, based on which the selection guidance of these two observers is indicated.