Fault Detection And Estimation And Fault Tolerant Control For Multi-Agent Systems

In recent years,with the development of science and technology,the control systems are becoming widely popular in human'various fields.The reliability and safety of control system has received great attention due to the increasing complexity of modern control systems.Fault tolerant control has been developed to guarantee the safe operation of the systems and to matain the acceptable system performance once faults occurred on actuators,sensors or other system components.At present,with the rapid development of multi-agent systems in the fields of aerospace,industry and agriculture,the research on multi-agent systems has been deepened gradually.More attention has been paid to fault tolerant control due to the large distribution of system components and the increasing probability of faults in multi-agent systems.In this paper,the problem of fault diagnosis and fault tolerant control is investigated for multi-agent systems.Based on the design methods of traditional observers,the distributed observer is constructed by using the neighborhood output estimation error,and the corresponding fault detection and estimation scheme is proposed.Based on neural networks,adaptive control,backstepping control and other technologies,the corresponding adaptive fault tolerant control schemes are proposed to compensate for the faults' effect on system performance.The main research work in this thesis is as follows:Firstly,the adaptive fauit detection and estimation problems are investigated for a class of linear multi-agent systems with sensor faults,and adaptive neighborhood-observer-based approaches are respectively proposed to detect and estimate the faults Based on graph and Lyapunov stability theory,it is proved that the detection and estimation errors converge to an adjustable small neighborhood of the origin with all signals in the closed-loop system being bounded.Finally,simulation results show the effectiveness of the scheme proposed in this paper.Secondly,the problem of distributed adaptive fault tolerant control is investigated for nonlinear multi-agent systems with sensor faults.By utilizing radial basis function neural networks to approximate the unknown continuous nonlinear functions in the system,a distributed-observer-based adaptive neural networks scheme is proposed to estimate each follower's unmeasured state.Then,by using the sliding mode control technique,a distributed adaptive controller is proposed for each follower.Based on the graph theory and Lyapunov stability theory,the stability of the closed-loop system is proved,and the tracking errors converge to a small adjustable neighborhood of the origin.Finally,simulation results are given to demonstrate the effectiveness of the proposed control scheme.Thirdly,the problem of distributed adaptive output-feedback fault tolerant control is studied for multiagent systems.Firstly,under the condition that the input signals of radial basis function neural networks deviate from the true values duo to sensor faults,the approximation capability of the neural networks is investigated.By analyzing the structrer of the neural networks,the analytical expression of output layer of RBF neural network under the condition of input layer signal distortion is obtained.Then,combining the filter observer and the adaptive backstepping technique,a distributed adaptive neural output-feedback control scheme is proposed to guarantee the output consensus of all nodes under undirected communication graphs.Based on the graph theory and Lyapunov stability theory,it is proved that the proposed adaptive neural control scheme guarantees the uniformly ultimate boundness of the closed-loop systems,and the tracking errors converge to a small adjustable neighborhood of the origin.The simulation results demonstrate the effectiveness of the proposed control scheme.Frouthly,the active adaptive fault tolerant neural control problem is discussed for large-scale uncertain systems against actuator faults.The unknown interconnections among subsystems are assumed to be nonlinear,not traditional linear.A general actuator fault model is proposed,which integrates bias and gain time-varying faults.Then,based on Lyapunov stability theory,a novel fault diagnostic algorithm and accommodation scheme are proposed,where the assumptions in the existing works are removed and fault tolerant controller singularity problem is avoided.Finally,simulation results show the efficiency of the presented control approach.