Research On H_? Fault Detection Methods For Complex Stochastic Nonlinearity Systems

In recent decades,with the development of fault detection technology,traditional point-to-point control systems are not suitable for control ideology and performance requirements of the modern control.Therefore,the network control systems have received more and more research attention and are widely used in engineering practice.However,the introduction of network also leads to a series of network induction phenomena,such as channel fading,time-varying delay and random occurrence of nonlinear,etc.The H_? filtering technology can not only ensure the stability of the global system under the premise of these disturbances,but also can make the filtering dynamic system have a certain H_? performance.Accordingly,the H_? filtering research based on the network control system has important engineering significance,which is also the work to be deal with in this paper.The filtering of a series of complex systems with random nonlinearity is discussed in this paper,and the matching fault detection problem is also solved.The contributions of this paper are as follows:Firstly,considering several kinds of network control systems,including random systems,nonlinear systems,fuzzy systems,time-varying delay systems,etc.,the matching system model is established.Integrated with some current hot research areas,such as random occurrence of nonlinear,event triggering mechanism,stochastic communication protocol and other network inducement factors,the corresponding fault detection filter is projected.Considering the influence of channel fading,the expected H_? performance constraint is given beforehand,and the homologous non-fragile H_? filtering problems concerned the above kinds of network control systems are researched.Secondly,the nature of the aforementioned network induced factors is richly studied and thses factors are modeled mathematically.The white sequence of Bernoulli distribution is used to describe the nonlinear disturbance appearing in the fault detection filtering in a random manner.Using a modified rice fading model to reflect the measurement signal received by the fault detection filter,the channel coefficient has an arbitrary probability density function on the interval [0,1].The event controller is used to reduce unnecessary data transmission in the communication channel.Until the pre-specified "event" is triggered,the current measurement signal is sent to the fault detection filter.The stochastic communication protocol is used to randomly determine which sensor will have access to the communication network at a certain time,and the markov chain is used to describe the random communication probability model.Using fuzziness and fuzzy membership functions to fit a special nonlinearity at the level of network control system,the expert experience can be better integrated into the control system.The actual parameter variation is described by the multiplication gain variations,which realizes the non-fragile of the fault detection filtet.Combining the above system initial state with the corresponding filter model,the filter dynamic error augmented system is established to facilitate the next analysis.Finally,Based on Lyapunov stability theory,stochastic analysis technology,linear matrix inequalities skills along with H_? filtering technology,the original fault detection problem is converted into an equivalent H_? filtering problem,which greatly reduces the algorithm complexity.The sufficient conditions for the existence of the fault detection filter based on the studied several types of the network control systems are established,which not only guarantees the stochastic stability of the filtering error system,but also makes the designed fault detection filter have a certain H_? performance.The gain of the addressed fault detection filter can be obtained by solving the feasible solution of a convex optimization LMI.The validity of the algorithm is verified in the numerical simulation examples at the end of each chapter.