H_∞ State Estimation For Complex Networks:the Partial-Nodes-Based Case

For the past few years,the complex networks have obtained continuous attention from researchers due to their wide applications in many actual systems such as Internet,power networks,communications networks,social networks,and so on.Typical complex networks are composed of a lot of nodes that are topologically connected with each other,and each node can be regarded as a basic unit with specific content and rich information interaction behaviour.Complex networks exhibit rich dynamic behaviors,which have attracted extensive attention from various fields.In particular,a lot of research has been done on the synchronization,consensus,stability analysis and pinning control of various complex networks.The state information plays a key role in the research of dynamic behavior of complex networks.Sensors can be used to directly measure the state information of the complex networks with a small number of nodes.But many actual systems have a large scale.For example,the power networks have thousands of nodes.It is usually difficult to get state information of power networks due to its large-scale.In this case,how to estimate the state information of the complex networks has become a focus of research,which leads to the state estimation problems.This thesis mainly studies the problem of H_∞state estimation to ensure that the estimation performance meets the H_∞performance.In the existing state estimator research,the measurement outputs are usually collected from all the network nodes.In fact,due to the large scale of the complex networks and the interaction between nodes,the measurement outputs of some nodes may be redundant and may contribute little to the estimator.In this case,it is not necessary to utilize measurements from all network nodes to complete the status estimation task when resources become a concern.However,there are very few results about the problem of partial-node-based state estimation,and this is one of the main motivations of this thesis.In actual engineering,the sensor is often saturated as a result of physical constraints,and the channel fading phenomenon will inevitably occur in the signal transmission process between the sensor and the estimator.What is particularly noteworthy is that sensor saturation and channel fading,if not considered,would degrade the estimator performance.In view of this engineering reality,this thesis extends some existing research results to the problem of H_∞state estimation based on partial nodes.For a class of discrete nonlinear complex networks with sensor saturation and channel fading,a partial-node-based H_∞state estimation is designed.The framework of this thesis generally includes the following chapters:In Chapter 1 and Chapter 2,the research background,motivation and some related theoretical are discussed.Chapter 3 address the partial-node-based H_∞state estimation problem for a class of discrete-time nonlinear complex networks with sensor saturations and fading channels.The influence of sensor saturation and channel fading on the estimation accuracy is considered.Finally,we design a partial-node-based H_∞state estimator,and its theoretical correctness is proved by theoretical derivation.In chapter 4,a simulation example is provided to shown the availability of the obtained partial-node-based H_∞state estimator.The gain matrix is calculated according to the linear matrix inequality method.The comparison between the estimated value and the actual value in the simulation result verifies the effectiveness of the estimator designed in this thesis.