Filtering Algorithms For Nonlimear Networked Control Systems With Random Delay And Packet Dropout

In the control problems of systems,the state estimation has always been one of the mainstream research issues.With the rapid development of internet technology,the crossintegration between internet technology and filtering technology is becoming larger and larger.While studying the problem of systems state estimation,we should also consider certain network phenomena will occur randomly in the systems.Therefore,it is worth discussing how to solve the filtering problem of control system with network phenomenon.Based on the previous research results,this paper mainly discusses the filtering problem of nonlinear networked control systems with stochastic delay and packet dropout.The main content is divided into four parts.In the first part,a new unscented Kalman filtering algorithm is proposed for nonlinear stochastic systems with packet dropout compensation and correlated noises based on the unscented transformation approach.In the second part,a particle filtering algorithm is researched in the framework of Bayesian filtering for a nonlinear system with random time delay and missing measurements.In the third part,a hybrid Kalman filter is designed for the nonlinear system with multi-sensors by combining the linear filtering method with the unscented transformation method.In the fourth part,a hybrid Kalman filter with stochastic nonlinearity functions is constructed based on the background of maneuvering target tracking system,and an interactive multi-Kalman filtering algorithm is studied.The specific research contents of this paper are as follows:The problem of unscented Kalman filtering algorithm is studied for nonlinear discretetime systems with random packet dropout compensation.A random variable,which obeys Bernoulli distribution with known conditional probability,is introduced to depict the phenomenon of packet dropout occurring in a stochastic way.Here,we use one-step predicted value as a compensator to estimate the state of the system instead of 0 input.In the algorithm,we select two sigma point sets to approximately calculate the parameters of recursive unscented Kalman filter to improve the accuracy of the filtering algorithm.Under the principle of minimum system estimation error,a recursive Kalman filter is designed based on the unscented transformation method.Also,we consider the unscented Kalman filtering problem for nonlinear discrete-time system with correlated noises and missing measurements.By using the projective theory and the unscented transformation method,a one-step predictor is constructed to reduce the influence of system noises on the accuracy of filtering algorithm.Then,based on the one-step predictor,a recursive unscented Kalman filter is designed with correlated noise and missing measurements.The particle filter algorithm is researched for discrete-time nonlinear systems with stochastic delay and missing measurements.We need to introduce several random variables obeying Bernoulli distribution to characterize the phenomenon of stochastic multistep delay and missing measurements.Due to the excessive number of random variables,it is not easy to analyze and compare the effects of different delay rates and missing rates on the estimation performance of filter.Therefore,in order to facilitate the comparison,we take a nonlinear system with random one-step delay and missing measurements as an example to study.In the system model,two random variables satisfying Bernoulli distribution are introduced to characterize the phenomenon of sensor random time delay and missing measurements.Assuming that the system satisfies the first-order Markovian process,a recursive formula for calculating the importance weight of sampling is given in the framework of Baysian filtering,which can reduce the influence of random delay and missing measurement on the performance of the system filter and improve the accuracy and effectiveness of the filtering algorithm for system state estimation.In the numerical example,we compare our filtering algorithm with the traditional particle filter algorithm,and analyze the influence of different delay rates and missing rates on the performance of the system estimator.Considering the nonlinear system with random multi-step time delay and missing measurement,the recursive formula for calculating the weight of particle importance is given.An example is given to verify the accuracy of the multi-step delay filtering algorithm.The hybrid Kalman filtering algorithm is discussed for nonlinear discrete-time systems with multiple sensors.A diagonal matrix is introduced to describe the multiple missing measurements in the system.Each element of the diagonal matrix is a random variable satisfying Bernoulli distribution.A new hybrid Kalman filter is designed to solve the problem of state estimation for nonlinear systems with nonlinear stochastic functions by combining the derivation method of linear filtering(recursive projective formula)with the unscented transformation method.The hybrid filtering algorithm,which designed by us,not only reduces the influence of random missing measurements of multiple sensors on the performance of filter,but also can estimate the state of the hybrid system more accurately.Then,the consensus-based hybrid Kalman filtering is considered for nonlinear discretetime systems with multiplicative noises.In the system model,the multiplicative noises of the system are characterized by the random variables of zero mean and unit variance.By using the recursive projective formula and the unscented transformation method,a new hybrid Kalman filter with multiplicative noise is designed.And then,based on the consensus on information approach,a consensus algorithm based on hybrid Kalman filtering is presented.Through the simulation example,we can see that the consensus algorithm can make the state estimation results of multiply sensors unanimous and improve the effectiveness of multiply filters.In the background of maneuvering target tracking system,a hybrid Kalman filtering algorithm is proposed with stochastic non-linear function and packet dropout compensation.The establishment process of the system model is introduced.Based on the model of maneuvering target tracking system,we introduce stochastic nonlinear functions to describe uncertain disturbances in the systems.Based on the previous theoretical researches,the corresponding recursive hybrid Kalman filter is designed and applied to maneuvering target tracking system.Then,the interactive multi-Kalman filtering algorithm is considered.In practice problems,we can't utilize a system model to accurately depict the moving state of the target.Hence,we introduce an interactive multi-Kalman filtering algorithm.For the above maneuvering target tracking system,the position information of the target is estimated by using the interactive multi-Kalman filtering algorithm.The accuracy and effectiveness of the proposed filtering algorithm are verified by numerical simulation.