Studies On Control Methods Of The Uncertain System Subjected To Faults And Its Application

Modern aerospace vehicles can be affected by extremely challenging uncertainties and faults due to the critical flight conditions.Above all,highly nonlinear characteristics,actuator/sensor fault danger and complex coupling effects,combined with unknown multiple uncertainties cause a considerable challenge in designing control system for such kind of uncertain system,i.e.hypersonic reentry vehicles.A possible solution to maintain a high level of safety for fights control system gives raise to develop the Fault-Tolerant Control(FTC)strategies.In the face of highly uncertain nature and various faults risk of flight,many researchers have focused on the development of the methodologies for robust and reliable control systems,among which model-based fault diagnosis and identification,fault estimation(FE)and active fault tolerant control(AFTC)has many advantages due to its high fault-tolerant ability.Despite the high number of published works using this interesting framework,there still remains certain open problems in this field that are of great theoretical and practical interest.This dissertation,based on adaptive disturbance observer,sliding mode algorithm,robust control and control allocation,makes a deep study on model-based FE and AFTC,with the aim of developing an integrated fault estimation and fault tolerant control and analysis framework for uncertain systems subjected faults/failures.First of all,the design of sliding mode observer for unknown state and fault estimation of the uncertain system is studied.A smooth sliding mode(SSM)algorithm is first designed to provide reliable estimation under a smooth disturbance assumption.This assumption is next relaxed with the second proposed adaptive sliding mode(ASM)algorithm that deals with disturbances of unknown bounds.In addition,the analysis of the observers are extended to the case where measurements are corrupted by bias and noise.The gains of the proposed algorithms were deduced from the Lyapunov function.Furthermore,some useful guidelines are provided for the selection of the observer turning parameters.Then,the proposed sliding modes observation techniques are used for attitude estimation and fault detection problems for Unmanned Aerial Vehicles.A two-step procedure is proposed to improve attitude estimation in the presence of IMU biases and measurement noise.Using a quadrotor platform,we have demonstrated the potential performance of the proposed solution.Secondly,we concentrated upon actuator fault estimation based on sliding mode algorithms.Consider the observer matching condition that the product of the output and fault matrices must have full column rank is not satisfied,a novel sliding mode observer is constructed by using a compensated observer or high order differentiator to extend the system outputs.Generating the estimation of faults in finite time was possible via utilizing a super-twisting sliding mode observer,which also leads to an almost decoupled FE design from system uncertainties with great design freedom.Thirdly,this paper presents an adaptive FTC strategy to accommodate the faults of actuators by integrating a FE module within an adaptive FTC scheme.The proposed solution is designed to be robust against imperfections due to the inherent interactions between them.A sliding mode observer is designed to estimate the loss of effectiveness of the actuators,uncertainties and external disturbances,whose time-varying rate may be high but bounded.Signals provided by the FE module are next used by an augmented fault-tolerant control allocation scheme to accommodate the fault.More precisely,the tracking problem has been addressed with a robust adaptive model-reference Integral Sliding Mode(ISM)control law in a state feedback setting.Next,a closed-loop performance analysis is performed to know the impact of observer and controller imperfections on the control performances in spite of the presence of faults and perturbations.Fourthly,in order to improve superior transient performances for state tracking,the proposed method relies on a sliding mode controller combined with an Adaptive Disturbance Observer(ADO)and Reference Vector Generator(RVG).Firstly,an adaptive disturbance observeris designed to estimate the system perturbations with overshoot attenuation.Then,a multi-variable super-twisting sliding mode controller is developed combined with a reference vector generator together with the estimated perturbations,which improves superior transient performances for state tracking.This structure allows for a faster response and reducing the overshoots.Robust stability and performance guarantees of the overall closed-loop system are obtained using Lyapunov theory.Simulation results obtained on the nonlinear model of hypersonic vehicle illustrates the benefit of the proposed techniques that allows a faster response and minimizing the overshoots compared to linear conventional Disturbance Observers(DO)based sliding mode controller.