Optimized Distributed Filtering For Several Classes Of Saturated Nonlinear Systems With Communication Constraints

The issue of distributed filtering for nonlinear stochastic systems is one of the hottest topics in the field of control.Accordingly,the related filtering approaches have been extensively used in various domains such as power systems,military systems and industrial systems.With the rapid developments of communication technology,the network has played an important role in industrial applications and people's daily life.Although it has the merits of resource sharing,high reliability and strong flexibility,the injection of communication network induces a verity of communication limitation phenomena,such as bandwidth limitations,missing measurements,cyber attacks,outliers and so on,which leads to the case that the remote filter side can only receive the incomplete measurements.In this thesis,the optimized distributed filter is constructed for several classes of saturated nonlinear systems with communication constraints based on the variance-constrained method,and the performance of the filtering algorithm is analyzed.The specific research contents are summarized as follows:Firstly,the problem of distributed filtering is investigated for time-varying saturated nonlinear system with missing measurements under stochastic communication protocol.A sequence of random variables with uncertain occurring probability is employed to describe the missing measurements,and the uncertain occurring probability is provided according to the nominal mean and the upper bound of probability error.Moreover,the stochastic communication protocol is utilized to determine the network permission of nodes at each time instant.A novel optimized distributed filter is established based on the measurable information.The upper bound of the filtering error covariance is proposed by virtue of the proper inequality method,and a suitable filter gain is designed to optimize the upper bound.Then,a sufficient condition is introduced to ensure the uniform boundedness of the trace of the upper bound.Furthermore,the impact between the missing probabilities and the distributed filtering algorithm performance is analyzed accordingly.Finally,a numerical simulation is performed by Matlab to testify the superiority of the presented optimized distributed filtering algorithm.Secondly,the problem of optimized distributed filtering is discussed for discrete delayed saturated nonlinear system under random occurring cyber attacks.A set of random variables obeying Bernoulli distribution is adopted to characterize the cyber attacks,which is used to realize the conversion between different attack types.Besides,it is assumed that the occurrence probability of the attacks is permitted to be uncertain.A time-varying distributed filter is designed with the aid of the measurable information in the remote filter side.Moreover,the specific expression form of the upper bound of the filtering error covariance is received via the stochastic analysis method,and the filter parameter is designed to optimize the obtained upper bound based on the matrix simplification technique.In addition,a sufficient condition is proposed to assure the uniform boundedness of the trace of upper bound.Finally,the viability of the presented optimized distributed filtering strategy is demonstrated by a simulation example.Thirdly,the optimized distributed filtering issue is discussed for delayed saturated nonlinear systems with outliers and multiplicative noises.The X~2 detector is utilized to detect outliers,remove outliers and provide the compensation criteria.In the remote filter side,a novel distributed filter with compensation mechanism is designed.Next,the upper bound of the filtering error covariance is given in view of the variance-constrained method,and the proper filter gain is chosen to minimize the obtained upper bound.Besides,the optimized distributed filtering algorithm is designed via the compensation mechanism.Additionally,a sufficient condition is provided to guarantee that the trace of upper bound of the filtering error covariance is uniformly bounded.Finally,a numerical simulation is presented to exhibit the feasibility of the developed optimized distributed filtering algorithm.