Research On Distributed Estimation Methods Over Sensor Networks Under Different Communication Mechanisms

In practical systems,the states of the concerned plant always cannot be acquired directly,which only can be estimated from the measurement signals.On the other hand,sometimes the plant to be estimated is too complex and large,the traditional centralized estimation method presents poor fault tolerance and it is also hard to cover a large area,which cannot meet the actual requirements of the engineering.At present,the distributed estimation method over sensor networks has attracted extensive research attention of the scholars due to its adaptability,large-scale and reliability.In sensor networks,the sensor nodes transmit information through the networks.It is worth pointing out that the energy of the sensors and the bandwidth of the networks are always limited,which restrict the communications among the sensor nodes severely,in turn,poses a great challenge on the distributed estimation problem over sensor networks.In order to make full use of the valuable network resources,a variety of communication mechanisms have been proposed and introduced in the sensor networks to schedule the information transmission of the nodes.Naturally,the research on the distributed estimation over sensor networks under the influence of the communication mechanisms has become a very meaningful and challenging topic.In this paper,under different communication mechanisms,the distributed estimators are designed for several types of discrete time-varying systems by using Lyapunov stability theory,stochastic analysis method,matrix theory and other technologies.We are committed to designing a suitable estimator such that the estimation error meets the expected performance index on the premise of saving resources.Specifically,according to different communication mechanisms,this paper can be divided into the following parts:First,with the consideration of the Round-Robin protocol,the H_?distributed state estimation problem is settled for a class of discrete time-varying nonlinear systems over the sensor networks with the randomly occurring switching topologies.According to the Round-Robin protocol,one node is selected at each time instant which has the data transmission permission.The randomly occurring switching topologies are governed by random variables that obey Bernoulli distributions.The sufficient conditions are obtained for the existence of the estimators that guarantee the given H_?performance index of the estimation error dynamics,and the state estimator parameters are acquired by solving certain matrix inequalities.Second,considering the static event-triggered mechanism,the finite-horizon H_?distributed state estimator is carried out for a class of discrete time-varying systems with redundant channels.For the sake of reducing the packet drop rate,several transmission channels are utilized to improve the reliability of the data transmission.The static event-triggering condition is utilized to determine whether the data is sent or not,which can reduce the data transmission frequency in the network so as to achieve the purpose of reducing the bandwidth occupation.For the sensor networks with fixed and switched topology,the desired estimators that can satisfy the prescribed performance index are respectively designed.With the help of the Lyapunov stability theory,the sufficient criteria are acquired which can ensure that the estimation error dynamics achieves the given H_?performance index.Furthermore,the estimator parameters are acquired by solving a set of convex optimization problems.Then,under the dynamic event-triggered mechanism,the security-guaranteed distributed state estimation problem is investigated with the measurements only from partial sensors in the presence of the randomly occurring deception attacks.Moreover,the H_?distributed state estimation problem is dealt with for a class of discrete time-varying systems with randomly occurring parameter uncertainties over randomly switched topological sensor networks.In order to adaptively adjust the threshold of the event triggering condition and further optimize the event-triggering mechanism,a dynamic event-triggering mechanism is introduced whose triggering threshold is updated by a first-order autoregressive equation.The Kronecker function and Markov jumping parameters are employed to describe the random changes of the topology of sensor networks.Through the Lyapunov method and recursive inequality technique,a distributed state estimation method is proposed to guarantee that the overall state estimation error dynamics satisfies the given security performance constraints and the H_?performance index.The gain parameters of distributed state estimator at each time step are calculated by solving a set of recursive matrix inequalities.Finally,according to the research results,the event-triggering mechanism based distributed estimation problem over sensor networks with switching topologies are discussed for the dynamic positioning system of offshore drilling platform.In this chapter,the modeling method of event-triggering mechanism is the same as that in Chapter 3.First of all,the motion model of the offshore drilling platform is constructed according to the motion of the offshore drilling platform in the ocean,and then the distributed estimator is designed which satisfies the expected performance index.Then,the proposed estimator is applied to the estimating of the dynamic positioning system of the offshore drilling platform,and the effectiveness of the method is verified by a simulation.