Research On Set Membership Filtering Of Networked Discrete System Under Communication Protocols

A networked control system(NCS)is a control system constituted by a network.The system establishes a communication channel between physical spaces through the network,so it can perform information transmission and information processing over long distances.In this thesis,we studied under the network environment set membership filtering problem of discrete time varying system.The application of the network environment will lead to network induced phenomena(such as time delay,state saturation and network attacks,etc.).We introduce some communication protocols(such as round-robin(RR)protocol,stochastic communication protocol(SCP),etc)to achieve the goal of reducing network induced phenomena.For the problem of discrete time-varying set membership filtering of nonlinear systems,the main purpose of this thesis is to design the networked system under different communication protocols set membership filter,in the meanwhile discuss the effects of time delay,incomplete measurement information,bounded noise,and network attacks on filtering performance.The main contents are as follows:The problem of set-membership filtering under the RR protocol is addressed for a category of time delay discrete time-varying networked systems.Considering that there is state saturation in the system and that the noise in the system is unknown but bounded,and applying the RR protocol to transmit data in order to save communication network resources,a set membership filter is designed under the incompleteness of the measurement information.Using the LMI method,establishing a sufficient condition to ensure that the filtering error can be restricted to the specified set of ellipsoids,and the optimization algorithm of the ellipsoid where the filtering error is located is given.By solving the optimization problem,the filter gain matrix that minimizes the filtering error is obtained.Numerical examples are used to verify the feasibility of the proposed filtering scheme.To address the problem of probability-guaranteed distributed set membership filtering under the SCP for a class of time-delay discrete time-varying sensor networked,the sensor saturation distributed delay and the uniform quantization are considered.Apart from this,there is also the time delay which is distributed time delay.The SCP is introduced to adjust the data transmission between sensor nodes and the measurement output of the system is quantized,which to achieve the purpose of ease to limited network bandwidth.The measurement information of sensor saturation and quantified is obtained,these measurements are used to design a set membership filter.Then,using the LMI method,a sufficient condition is given to make sure that the filtering error can be bounded to the specified set of ellipsoids.,and the optimization algorithm for the ellipsoid where the filtering error is located is given,the filtering error is minimized by solving the optimization problem.Simultaneously,get the filter gain.A numerical example is presented to test the feasibility of the proposed filtering scheme.An exploration of probability-guaranteed distributed set membership filtering for a class of nonlinear time-varying systems with distributed delays,injected attacks and sensor saturation under sensor networks.The system has both random noise and bounded noise interference.In a sensor network,the measurement output of the system is affected by sensor saturation and injection attacks.Assuming that the attack signal is unknown and restricted to a certain ellipsoid set,SCP is used to randomly adjust the data transmission between sensor nodes.Design a probability-guaranteed distributed set membership filter under injection attack.Through the LMI method,a sufficient condition to ensure that the filtering error is limited to a given ellipsoid set with a predetermined probability is given,and the convex problem of optimisation is solved to obtain the optimal filter gain matrix.Finally,the feasibility of the suggested filtering method is demonstrated by numerical arithmetic examples.