Studies On Filtering Problems For Networked Multi-Rate Systems With Communication Protocols

Recently,with the fast development of the communication technology,the net-worked systems have been widely applied in many areas.In the networked systems,to alleviate the data collisions,communication protocols are often used to schedule the data transmissions among the system components.On the other hand,the multi-rate sampling strategy is often used in the real-world networked systems to achieve a satisfactory balance between the system performance and the resource consump-tion.Up to now,the studies on the filtering problems for networked systems are mainly focus on the single-rate systems,and the influence of the communication pro-tocols does not receive adequate research attention.As such,it is a meaningful and challenging topic to develop filtering theory for networked multi-rate systems with communication protocols.In this thesis,the filtering problems are studied for several classes of networked multi-rate systems,such as multi-rate time-varying systems,multi-rate time-delay systems,and multi-rate linear repetitive processes,with the communication proto-cols.To solve the filtering problems,some novel models and novel methods are proposed.For example,a novel model is proposed to describe the measurement after scheduled by the Round-Robin protocol,a novel selection principle of the weighted Try-Once-Discard protocol is proposed to adapt the asynchronous sensors,and a novel method is proposed to convert the multi-rate systems to single-rate ones.The content of this thesis can be divided into five parts.In the first part,the research background and the motivation of this thesis are discussed,and the research content and the contributions of this thesis are concluded.In the second part,to facilitate the study of the filtering problems for networked multi-rate systems with communication protocols,the filtering problems for net-worked multi-rate systems are investigated.First,the H_?state estimation problem is studied for multi-rate artificial neural networks with integral measurements.To take the multi-rate sampling into account,the state update period of the artificial neural networks is allowed to be different from the sampling period of the sensor.A measurement model is established to describe the phenomenon of integral mea-surements.A novel method is proposed to convert the multi-rate artificial neural networks to single-rate ones.With the help of the Lyapunov-Krasovskii functional,the analysis results on the existence of the H_?state estimator are derived and the desired H_?state estimator are designed.Then,the non-fragile H_?filtering problem is investigated for multi-rate time-delayed systems with missing measurements over sensor networks.In order to take the multi-rate sampling into account,the state update period of the system and the sampling period of the sensors are allowed to be different.The variation of the filter gains is considered by introducing uncertainty terms in the filter gains.The probabilistic packet dropout that occurs during the information transmissions among the sensor nodes is characterized by the Gilbert-Elliott model and a novel measurement model is established.With the help of the Lyapunov-Krasovskii functional,the existence condition of the H_?filter is derived and the non-fragile distributed H_?filter is then designed.In the third part,based on the results obtained in the second part,the filter-ing problems are studied for the networked multi-rate systems with Round-Robin protocol.First,the l₂-l_?state estimation problem is concerned for the delayed ar-tificial neural networks under high-rate communication networks with Round-Robin protocol.The high-rate communication network is taken into consideration and a novel model is established to describe the measurement after transmitted through the high-rate communication networks with Round-Robin protocol.The existence of the desired state estimator is analyzed by constructing the Lyapunov-Krasovskii functional and the l₂-l_?state estimator is designed.Then,the recursive state esti-mation problem is studied for the multi-rate time-delayed systems with Round-Robin protocol.For the considered system with stochastic nonlinearities and distributed time-delays,an iterative method is presented to transform the filtering problem for multi-rate system into filtering problem for single-rate system.A novel model is provided to describe the measurement scheduled by the Round-Robin protocol.By solving certain matrix difference equations,the minimum upper bound of the estima-tion error covariance is derived and the gain of the state estimator is characterized.In the fourth part,the filtering problems are studied for the networked multi-rate systems with weighted Try-Once-Discard protocol.First,the recursive filtering prob-lem is studied for multi-rate time-varying systems with measurement outliers under the weighted Try-Once-Discard protocol.The multi-rate sampling,the measurement outliers,and the scheduling effect of the weighted Try-Once-Discard protocol are considered.The multi-rate system is converted to a single-rate one by using the lifting technique,and a mathematical model is given to describe the measurement after scheduled by the weighted Try-Once-Discard protocol.A novel filter structure is proposed which is able to effectively restrain the influence of the measurement out-liers on the filtering performance.Based on the solution to certain matrix difference equation,the filter gain with the recursive form is derived.Furthermore,the analysis result is derived on the exponential boundedness of the filtering error dynamic in the mean square sense.Then,the recursive fusion estimation problem is studied for the multi-rate linear repetitive processes with weighted Try-Once-Discard protocol.The sampling periods of the sensors are allowed to be different,and the state update pe-riod of the linear repetitive processes is also allowed to be different from the sampling periods of the asynchronous sensors.A novel selection rule of the weighted Try-Once-Discard protocol is designed to adapt the asynchronous sensors.The local recursive state estimators are first designed by solving matrix difference equations,and then the local estimates are fused by recurring to the sequential covariance intersection fusion method.In the last part,the H_?filtering problem is investigated for multi-rate systems with randomly occurring sensor saturations under the p-persistent CSMA proto-col.The sensors have different sampling periods which are also different from the state update period of the system.A set of mutually independent Bernoulli dis-tributed white sequences is introduced to characterize the random occurrence of the sensor saturations.A novel analytical method is developed to deal with the time-varying probability issue caused by the coupling of the multi-rate sampling and the p-persistent CSMA protocol.By using the lifting technique,the filtering problem for multi-rate system is converted to the filtering problem for single-rate system.By constructing the Lyapunov function,sufficient conditions are established on the exponential mean-square stability and the H_?performance of the filtering error dy-namics,and the H_?filter is designed by solving the matrix inequality.Finally,the proposed filtering scheme is applied to the continuous stirred tank reactor to verify the effectiveness and practical significance of our result.