Filtering And Control Of Networked Control Systems Under Probability Constraints

The research of networked control systems involves many fields of electronics,communication and control.The causal relationship between "sensing-control-drive-controlled" is extremely complicated.When there are constraints on the physical variables of the system,how to reasonably establish the model,analyze and solve the filter/controller design problem is a challenging research work.In addition,there are still many interference factors in the actual networked control system.There are many challenging research work to do,such as how to accurately model the system to describe the internal mechanism of the system how to design a reasonable filter/controller to make the system meet the constraints and meet the requirements expected performance.In practical systems,the physical variables of the system often have various variable constraints.The probabilistic constraints studied in this paper are one of the variables constraints.Probabilistic constraints refer to system variables satisfying constraints or constrained regions with a certain probability.In other words,system variables can violate constraints or deviate from constrained regions with a small probability.In this paper,a filtering/control method based on probability constraints is proposed.Considering the successive packet dropouts,sensor saturation,multiplicative noise and stochastic nonlinearity of networked control system,a new networked control system model is established,and the corresponding filtering and tracking control problems are analyzed and solved.Firstly,based on the research of probability constraint,this paper makes an in-depth study on the tracking control problem of networked control system based on event triggering method.The multi-dimensional Chebyshev boundary is used to deal with the conditional transformation of probability constrained ellipsoid.Considering the existence of random noise and multiplicative noise in the network environment,a new networked control system and filter model are established.Then,by stochastic analysis method and linear matrix inequality analysis method,the sufficient conditions for the existence of the optimal filter of networked control system and the design method of the filter are obtained.Finally,a simulation example is given to verify the effectiveness of the proposed method.Secondly,the finite-time optimal filter design for a class of discrete time-varying systems with sensor saturation is studied.By adding stochastic nonlinearity,a new networked control system model is proposed.Sufficient conditions for the finite-time optimal filter to satisfy probability constraints are given,and the design algorithm of the optimal filter is given.Finally,an example is given to verify the effectiveness of the proposed method.Finally,the tracking control problem of networked control system is studied.Considering the multiplicative noise and stochastic nonlinearity in the network environment,a new networked control system model is obtained.On this basis,a variance constrained optimal estimator(filter)is designed.Using similar analysis methods,we can get the design method of tracking controller under probability constraints.So that we give a tracking controller its estimation error satisfies variance constraints and tracking error satisfies probability constraints.Finally,an example is given to verify the effectiveness of the proposed method.