Fault Diagnosis Of Multi-Agent System Based On Lyapunov Function

With the rapid development of modern control technology,the requirements for system stability and security are also increasing.Because of the uncertainty and nonlinearity in the actual control system,the stability analysis and fault estimation of nonlinear systems have attracted the attention of many scholars.The disturbance signal may have a great impact on the fault detection of the system,and even send an error warning signal.Therefore,when improving system stability and robust performance indicators,we must also detect faults in the system in time to avoid accidents.In this thesis,based on the state-dependent Lyapunov function,the stability analysis of nonlinear systems is carried out,and then the design method of fault estimation sliding mode observer for nonlinear multi-agent system is studied in the whole frequency domain,the main research work of this thesis is as follows:(1)A new method based on state-dependent Lyapunov function stability analysis is proposed for the stability analysis of nonlinear subsystems.Firstly,the Lyapunov function expression is constructed according to the nonlinear differential equation.Secondly,based on Lyapunov stability related lemma and variable gradient method,the analysis of LPV system is carried out,and the less conservative stability conditions in the form of linear matrix inequality are obtained.Then,the stability condition is extended to an LPV system with parameters that are linear combination of state variables and linear systems subject to actuator saturation.Finally,the effectiveness of the proposed algorithm is verified via five numerical example.(2)A fault estimation sliding mode observer is designed in the whole frequency domain for the case of actuator failure in a nonlinear multi-agent system.Firstly,taking the actuator faults as auxiliary states,an augmented system state equation is constructed.According to the Kronecker inner product property,the state equation of the nonlinear multi-agent system is obtained.Secondly,according to the information interaction between the agents in the directed graph,the relative output estimation error vector is introduced to established a fuzzy adaptive fault estimation sliding mode observer for a single agent.According to the general approximation theorem of fuzzy system,the theory of sliding mode observer and the inner product of Kronecker,the error dynamic system equation is obtained.Then,the robust performance index is introduced to obtain sufficient conditions for satisfying the H_?performance index under the linear matrix inequality.Finally,the fault estimation sliding mode observer designed can accurately estimate the fault signal by Matlab simulation.