Optimized Filtering And Its Application For Several Classes Of Nonlinear Stochastic Systems With Communication Constraints

The problem of state estimation or filtering for nonlinear dynamical systems is one of the hottest research issues.The corresponding estimation approaches are mainly used in power system,radar location,face recognition and so on.To study and analyze the performance of dynamical systems,we need to know the state information of the system.However,the state information of the system is often immeasurable,hence it is very important to design an effective state estimation/filtering algorithm to track the state of the dynamical system.On the one hand,the nonlinearity is ubiquitous in real systems.Hence,it is necessary to propose a suitable nonlinear state estimation/filtering method to weaken the effects induced by the nonlinearity.On the other hand,the remote state estimator/filter cannot receive the corresponding useful information in real time,but only part of information can be transmitted to the state estimator/filter due to the influence of communication constraints,such as signal quantization,data loss,communication protocol and state saturation,etc.Therefore,it is of great significance of theoretical research and practical value to estimate the unknown state of the system effectively based on available partial effective information.In this thesis,based on variance constraint approach,some new recursive state estimation/filtering algorithms are designed for nonlinear dynamical systems with communication constraints.The specific contents are given as follows:Firstly,the robust filtering problem is investigated for a class of time-varying nonlinear systems with stochastic occurring uncertainties and signal quantization.The modeling error is characterized by norm bounded uncertainty and the signal quantization is realized by a logarithmic quantizer.Based on the measurable information of the system,a novel filter is constructed.Then,the upper bound of filtering error covariance is given by means of mathematical methods.Next,an effective filter gain is designed to minimize such an upper bound.In addition,asufficient condition is given to guarantee the mean-square exponential boundedness of the filtering error.Finally,a Matlab simulation example is given to verify the superiority and effectiveness of the proposed filtering strategy.Secondly,based on the variance-constrained approach,the resilient filtering issue is discussed for a class of time-varying systems with stochastic uncertainties and successive packet dropouts.In order to reduce the burden of network transmission and save network resources,an event-based communication mechanism is employed.The stochastic uncertainties are discussed to characterize the modeling errors and a random variable obeying the Bernoulli distribution is used to describe the successive data loss situation.Based on the measurable information,an effective time-varying filter is constructed.Then,the upper bound of filtering error covariance is given by utilizing the inequality technique.In addition,we design the filter gain to minimize such an obtained upper bound.Finally,the effectiveness and feasibility of the proposed algorithm are verified by Matlab numerical simulation.Thirdly,the event-based resilient state estimation problem is addressed for a class of time-varying coupled complex networks with multiple missing measurements and randomly switching topologies.A set of random variables with deterministic probability distribution is employed to characterize the phenomenon of multiple missing measurements.An event-triggered communication mechanism is employed to reduce the possibility of data congestion.Meanwhile,some Bernoulli random variables and white noise are used to describe the randomly switching topologies behavior.Based on the measurable information of the system,a new time-varying resilient state estimator is constructed.Then,we give the specific expression of the upper bound of the estimation error covariance.Next,a reasonable estimator gain is designed to minimize the upper bound.Finally,the validity and feasibility of the developed algorithm are verified by Matlab numerical simulation.Fourthly,based on the variance-constrained method,the robust state estimation problem is studied for a class of time-varying state-saturated complex networks with random access protocol.The random access protocol is introduced to adjust the transmission order of information between network nodes so as to avoid the possible occurrence of the network-induced phenomenon.A time-varying robust stateestimator is constructed based on the measurable output information.Then,using the mathematical methods,we get the recursive equation of the upper bound of estimation error covariance.Next,an effective estimator gain is designed to ensure that the upper bound is optimized.In addition,the sufficient condition is given to ensure that the trace of upper bound is uniformly bounded.Finally,the validity and feasibility of the proposed algorithm are verified by Matlab simulation.