Non-fragile Joint State And Fault Estimation For Several Classes Of Time-varying Complex Networks

In the wake of developments in science and technology,more and more experts and researchers have paid close attention to complex networks with the Internet as a representative.Furthermore,the common complex networks,such as social network,biological network,electricity network and so on,are also more widely applied in real life.Hence,there is a need to master nodes' state information of complex networks,which contributes to have a better understanding of complex networks,makes the most of the advantages and rejects the disadvantages.However,it is almost impossible to obtain the state information of all nodes by direct measurements,due to a mass of nodes and complex structure of the complex networks.Therefore,it has a practical significance to estimate all nodes' state of the complex networks based on the available measurement information.It is worth mentioning that the state estimation problem for complex networks has made significant achievements.In contrast,there is still a great deal of research space on the non-fragile state estimation problem for complex networks.With the wide application of complex networks,if the system of complex networks breaks down,a lot of manpower and financial resources will be wasted.Thus,the fault estimation for complex network is necessary.This thesis mainly studies non-fragile joint state and fault estimation problem for several classes of time-varying complex networks.The research contents are listed from the following four aspects:Firstly,the non-fragile joint state and fault estimation problem is studied for complex networks with missing measurement.Two sets of random variables with known occurrence probabilities,which obey Bernoulli distribution,are used to characterize the phenomena of missing measurement and switching topology,respectively.A estimator is designed based on the available measurement output.The recursive upper bound of the estimation error covariance is obtained by utilizing the basic inequality and matrix theory.Next,the proper estimator gain is constructed to solve the optimal problem of the upper bound,and the criterion is given,which ensures the upper bound to be bounded.Lastly,the availability of the presented joint estimation algorithm is examined by numerical simulations.Secondly,the non-fragile joint state and fault estimation problem is discussed for a class of complex networks subject to randomly occurring nonlinearities.Three groups of random variables with uncertain probabilities are used to describe the phenomena of randomly occurring nonlinearities,missing measurement and switching topology.Based on the known output information,a new time-varying estimator is constructed.The gain perturbation of the estimator is characterized by a set of multiplicative noises.Based on inequality processing techniques,the upper bound of the estimation error covariance is derived,and a proper estimator gain,which is suitable for online implementation,is designed.Next,sufficient conditions are given to guarantee that the upper bound is bounded.Finally,the numerical simulations are provided to prove the feasibility of the estimation algorithm.Thirdly,the joint state and fault estimation problem is investigated for complex networks with uncertain inner coupling and time delay.A non-fragile and time-varying estimator is designed under the influence of uncertain inner coupling and time delay.By using the effective random analysis method,the covariance of the estimation error is calculated and its minimum upper bound is derived.The recursive from of the estimator gain is given.Next,criterions for boundness analysis are provided to discuss joint estimation algorithm performance.Lastly,the numerical simulation examples are given to demonstrate the availability of the proposed algorithm.Fourthly,based on measurable information of partial nodes,the joint state and fault estimation problem is settled for uncertain complex networks on the basis of variance constraint method.Based on the known measurement information,a new joint estimator of state and fault is designed.The uncertain terms in the estimation error covariance are handled by utilizing the matrix simplification technique.Hence,recursive equation to the upper bound of the estimation error covariance can be acquired.Then,the upper bound is optimized by designing a proper estimator gain.Finally,simulation experiments are provided to testify the feasibility of the optimal estimation algorithm.