Rolling Time Domain Estimation Of Networked Systems With Limited Measurement Output

In recent years,with the rapid development of communication,computer and control technology,networked control system has become a hot research in control theory.Although many fruitful results have been achieved,there are still many problems to be overcome.For example,when a control system transmits data through a shared network,data loss,quantization,measurement abnormality and other data constraints may occur,which makes system analysis and design more complex.Therefore,aiming at the above problems,this paper studies the controllability and observability of networked multi-agent systems,the moving horizon estimation of networked systems with random packet loss and quantization,and the moving horizon estimation of networked systems with abnormal measurements and random packet loss.The main contents of this paper are as follows:1)The controllability and observability of networked multi-agent systems are discussed.For two kinds of multi-agent systems,the relationship between external input feedback and internal neighbor information communication is given.In particular,the coupling relationship between the external input state and the follower neighbor information is given when the controllability of the two multi-agent systems can be equivalent.In addition,the local conditions for the controllability of the multi-agent systems are discussed,and then the relationship between the local controllability and the overall controllability of the multi-agent systems is obtained.Then,by analyzing the coupling relationship between the eigenvector of the system matrix and the corresponding elements of the leader node,the sufficient and necessary conditions of the system observability are obtained.Finally,the reliability of the conclusion is verified by model analysis.2)Moving horizon estimation algorithm for networked systems with random packet loss and logarithmic quantization is designed.Considering the quantization process described by Logarithmic quantization and the random packet loss process satisfying the Bernoulli distribution,a network system model with quantization and packet loss is established.Then,based on this model,a networked moving horizon estimation algorithm is designed.Finally,a simulation example is given to verify the superiority of the proposed estimation method.3)Moving horizon estimation of networked systems with abnormal measurements and random packet loss is studied.Considering that the abnormal measurement noise results in the abnormal measurement output and the random packet loss process satisfies the Bernoulli distribution,a networked system model is established.In addition,different from Kalman filter,a robust moving horizon estimation method is proposed,andsufficient conditions for the convergence of the estimator are given.Finally,a simple example is given to verify the superiority of this method compared with Kalman filter.