Robust Adaptive Fault-tolerant Control Of Systems With Actuator Faults

Practical engineering systems are inevitably non-linear subject to external disturbances and actuator failures.In most cases,system faults are events that happen often at unexpected moments of time and their further development into overall system failures may lead to consequences that take different forms and scales.Therefore,how to maintain stable and acceptable operation performance for a given dynamic system when a failure occurs is an important issue in control system design.As faults are difficult to foresee and prevent,fault-tolerant control(FTC)has been viewed as one of the most promising control technologies for maintaining certain specified safety-related performance of a life-critical system in the presence of unexpected actuator faults.This thesis adopts advanced control algorithms to address this problem from the following aspects:Considering the stabilization problem of dynamic systems with actuator faults and external disturbances.Robust adaptive fault-tolerant control algorithms are derived without the need for analytically estimating bound on actuator failure variables,and thus the resultant FDD-independent(Fault detection and diagnosis,FDD)control scheme has simpler structure and demands less online computations.It is shown that with the proposed control,both actuator and subsystem/component faults can be accommodated,and the state dependent growth disturbances can be effectively attenuated.The algorithm is validated via a formative mathematical analysis based on a Lyapunov approach and numerical simulations in the presence of external disturbances,parametric uncertainties,as well as severe actuator.Then,the results have been extended and inproved to uncertain stochastic systems subject to modeling uncertainties and actuator faults.The model-following control problem associated with a class of nonlinear systems in the presence of modelling uncertainties and actuator faults is investigated in the thesis.The particular interest lies in the development of designer-friendly and cost-effective control scheme.By combining model-reference mechanism with robust adaptive radial basis function(RBF)neural network(NN),several control algorithms are derived without the need for precise system parameters or analytical bound estimation on actuator fault variables.It is shown that the developed control algorithms are structurally simple and computationally inexpensive.An adaptive iterative learning reliable control(AILRC)strategy is developed for a class of nonlinearly parameterized systems subject to unknown time-varying state delays and input saturation as well as actuator faults.The novel data driven AILRC is constructed by a nonlinear feedback term and a robust term.With the resultant algorithms,the nonlinear influence brought by actuator faults,input saturation and state delays can be compensated.The convergence of the closed loop systems is proved through a new time-weighted Lyapunov-Krasovskii-Like composite energy function(CEF).For Multi-agent system,agent's actuator fault can cause severe performance deterioration,or even system instability,leading to catastrophic accidents.To achieve coordination of active loads in DC microgrids,two main control methods are proposed.As the actuator faults are known,a game-theoretic performance function is defined for each active loads.Then,a distributed control policy simultaneously minimizes all performance functions.A low-voltage dc microgrid,simulated in MATLAB/Simulink environment,is used to study the effectiveness of the proposed methodology.As the faults are unknown,a distributed control strategies based on robust adaptive method are designed to solve the coordination problem.