Research On Variance-Constrained Filtering For Stochastic Nonlinear Discrete Time-Varying Systems

The filtering problem for stochastic nonlinear discrete time-varying system is on important research topics in the estimation theory.The correlation estimation method is applied in many fields such as military,traffic and image processing.In order to analyze the performance of nonlinear discrete time-varying system,the state information of the system should be clarified.However,the state information of the system is usually not directly measured,so it is necessary to design an effective filtering algorithm to estimate the state of the dynamic system.On the other hand,the measurement information can not be completely transmitted to the filter end,due to the uncertainty,measurement delay,data loss and other factors in the process of network transmission.Therefore,it has practical value for the estimate of unknown state in the system by using some effective information.In this paper,we use variance constraint method to study the filtering problem for stochastic nonlinear discrete time-varying systems.The main contents are summarized as follows:1.The problem of filtering for discrete time-varying systems with stochastic nonlinearity and measurement delay has been studied.Two phenomena about network-induced,random occurrence nonlinearity and measurement delay,are considered in the system.In order to characterize the measurement delay and random occurrence nonlinearity,two series of random sequences subject to Bernoulli distribution are introduced in this paper.And the proposed system is extended when the measurement delay is handled.According to the measurement information received by the sensor termination,the corresponding filter is designed.Then the upper bound of the covariance matrix of filtering error can be obtained,by solving the Riccati-like difference equation,furthermore the trace of the upper bound is minimized by designing the corresponding filtering gain matrix.Finally,an example is given to verify the effectiveness of the filtering algorithm.2.Based on variance constraint,the filtering problem for discrete time-varying system with uncertainty and data loss is studied.The random sequence obeying Bernoulli distribution is used to describe the packet loss,and the norm bounded uncertainty is introduced to describe the modeling error of the system.At the same time,an event-triggered mechanism is introduced to improve the utilization rate of network resources.A robust state estimator is designed,by using the available measurement information.Based on the inequality processing technique,the upper bound of the estimation error covariance matrix is calculated,and designing the filter gain matrix under the optimization index.Finally,an example is given to illustrate the effectiveness and practicability of the algorithm be proposed.