Research On H_? Fault Detection For Nonlinear Stochastic Systems

In recent years,with the increasing demand of high performance,security and reliability requirements of dynamic systems,more research attention has been paid to the fault detection.Compared with the traditional point to point communication systems,many advantages have been brought to the networked control systems.However,phenomena have been occurred due to the introduction of the network,such as delays,nonlinearities,channel fadings,etc.,which has given the research of the networked control systems difficulties and challenges with a more practical significance.In this paper,by the Lyapunov function principle,LMI technology and stochastic analysis technology are utilized to model the complicated networked control systems,and the design method of stochastic nonlinear H_?fault detection filter is analyzed.The main work and contributions of this paper are as follows:Firstly,an L-th Rice fading model is established for the discrete system considering fading channels,and a kind of non-fragile filter is designed.The Lyapunov functions are selected,the stochastic analysis technique and Kronecker product are employed to calculate the mathematical expectations of the differential of functions.Then the cost function is acquired according to defining the filtering error system and the H_?performance index.By the calculation based on LMIs,the sufficient condition which satisfies the asymptotic stability of the error augmented system and H_?performance restraint is obtained,as well as the gains of the filter.LMI toolbox of the MATLAB software is utilized to figure out the feasible solution of the standard LMI problem,and the feasibility of the algorithm is verified through the simulation.Secondly,the H_?fault detection method for Markovian jump systems with time-varying delays and randomly occurring nonlinearities is studied.Bernoulli distributed white sequences are adopted to represent the phenomenon of randomly occurring nonlinearities in the network.According to the stochastic characteristic of multiple time-varying delays,an individual Bernoulli distribution is used to describe this phenomenon.Through the desired exponential stability and H_?disturbance attenuation,the gain characteristic of the fault detection filter is obtained.Thirdly,the stochastic nonlinear fault detection algorithm is used for time-varying Markovian jump systems.The nonlinear and multiple time-varying delays phenomena are described by the Bernoulli typed random variables,the fault estimator is designed and the corresponding algorithm is studied.By adopting the Lyapunov-Krasovakii functional and the stochastic analysis theory,the explicit expression of such estimators which satisfies the corresponding performance index is parameterized by employing a recursive linear matrix inequality approach.A simulation example is provided to verify the usefulness of the proposed methods.Finally,based on the wind turbine system model,the non-fragile H_?fault detection filter is designed by using the method of fault detection considering the influence of the uncertain factors,the performance of the system is described by the residual evaluation function.