Control,Filtering And Fusion For Sampled-Data Systems Under Communication Constraints

The sampled-data systems with periodic sampling,stochastic sampling,multirate sampling and event-triggering sampling have been widespread in practical engineering systems.Moreover,due mainly to the communication constraints including limited network bandwidth,unstable network connection and limited packet capacity,the phenomena such as data missing,saturation,quantization and fading channels often happen in networked systems,which jeopardize the systems' performances and bring challenges in analysis and synthesis problems for networked systems.Therefore,it is of great significance in both application prospect and theoretical value to consider the effects of sampling methods and communication constraints in control,filtering and fusion problems for sampled-data systems and develop control,filtering and fusion theory for sampled-data systems under communication constraints.In this thesis,the control,filtering and fusion problems are investigated for sampled-data systems with communication constraints.The content of this thesis is mainly divided into three parts.In the first part,we investigate the state-feedback control problem for the sampled-data system with noisy sampling intervals,and the effects of quantization and saturation are considered simultaneously.In the second part,the filtering and fusion problem are studied for sampled-data systems subject to sensors and relays with random transmission power,time-correlated fading channels and various sampling methods,some new recursive filters and fusion estimators are constructed.Based on filtering methods proposed in the second part,in the third part,we focus on the observer-based output-feedback control problem of sampled-data systems with various sampling methods.Meanwhile,we try to apply our theoretical research to the control of body slip angle of electric vehicles.The compendious frame and description of the thesis are given as follows.· The quantized/saturated control problem is investigated for sampled-data systems with noisy sampling intervals.The case that the sampling error of the sampled-data system follows Erlang distribution is considered.First,an equivalent discrete-time system is obtained for the continuous-time system.Then,a confluent Vandermonde matrix method is proposed to transform the matrix exponential into a tractable form.Finally,with the help of Lyapunov theory and linear matrix inequality technique,the quantized/saturated controllers are designed such that sampled-data systems are stochastically stable.· The robust recursive filtering problem is studied for sampled-data systems with amplify-and-forward(AF)relays.The parameter uncertainties are described by a set of norm-bounded matrices.An AF relay is adopted to forward the signal received by the sensor to the filter.A set of random variables with certain probability distribution is introduced to characterize the random transmission power of sensors and relays.By utilizing the average transmission power,a robust filter is first constructed.Then,an upper bound is recursively obtained for the filtering error covariance in the presence of random transmission power and parameter uncertainties.The desired gain matrix is further parameterized by minimizing the obtained upper bound.Moreover,the boundness stability is analyzed for the filtering error.· The robust recursive filtering problem is considered for sampled-data systems with time-correlated fading channels.The measurement received by the sensor is transmitted to the remote filter through the time-correlated fading channel where the channel coefficient evolves according to a certain dynamics and hence exhibits a time-correlated nature.By introducing a class of auxiliary variables,an augmented system is constructed to reflect the dynamics of the state and fading coefficient simultaneously.Then,a recursive filter is designed which is capable of online computation.Furthermore,an upper bound is guaranteed for the filtering error covariance for the possible parameter uncertainties as well as the time-correlated fading channels.With the help of completing-the-squares technique,filter gains are parameterized by minimizing the obtained upper bound.· The fusion estimation problem is addressed for stochastic uncertain sampleddata systems with time-correlated Rician fading channels and periodic sampling.A set of Gaussian distributed random variables is introduced to describe the parameter uncertainties.The sensor and the local filter communicate through a time-correlated fading channel where the channel coefficient shifts on the basis of a certain dynamic process disturbed by one-step correlated noises.To analyze the effects of time-correlated fading channels,a class of additional variables is first introduced by integrating the dynamics of channel coefficients and the state.Then,a new set of local filters is constructed and the unbiasedness of local filters is examined.Furthermore,the local filter gains are designed such that the local filtering error covariance is minimized.Subsequently,the cross-covariances among local estimates are computed.Based on unbiased local estimates,local filtering error covariances and cross-covariances,a fusion estimate is obtained by using the weighted least square fusion method.· The event-based fusion estimation problem is studied for uncertain multi-rate systems with stochastic nonlinearities and colored measurement noises.A new augmentation approach is proposed by which the multi-rate sampled-data system with stochastic nonlinearities and parameter uncertainties is transformed into the single-rate system.In order to eliminate the effect of the colored measurement noises,a measurement model with uncorrected noises is established.Based on the measurement model established,a set of local event-triggered filters is constructed and the upper bounds of the local filtering error covariances are obtained.By using the Lagrange multiplier method,the local filter parameters are designed such that the upper bound obtained is minimum.For the local state estimates,a new fusion estimation scheme is proposed with the help of covariance intersection(CI)method and the consistency of the proposed CI-based fusion estimation scheme is shown.· The finite horizon event-triggered control problem is studied for a class of timevarying systems with user datagram protocol(UDP)communication channels.For the sake of saving energy,an event-triggering mechanism is adopted to manage the signal transmission in sensor-to-estimator(S/E)channel and the UDP is employed in the unreliable controller-to-actuator(C/A)communication channel.The main purpose is to design an observer-based controller such that,when there exist the event-triggering mechanism and the UDP,the prespecified H?performance criterion is satisfied over a finite time-horizon.By employing the completing-the-square method and the backward Riccati difference equation technique,the existence conditions are first established for the desired event-based H?controller.Then,the controller gains are iteratively determined with the help of the Moore-Penrose pseudo inverse approach.· The event-triggered non-fragile H?control problem is investigated for the body slip angle of electric vehicles with onboard vision systems.Dynamics of electric vehicles with an onboard vision system is characterized by a linear model in which the body slip angle,the yaw rate,the lateral offset and the vehicle heading angle are chosen as state variables.To reduce the energy consumption in signal transmission,a dynamic event-triggering mechanism is adopted to manage the data transmission between sensors and vehicle controllers.The aim is to design a non-fragile event-triggered controller such that,for all possible parameter variations in controller gains,all the system states are asymptotically stable and the H?performance requirement is met.With the help of the Lyapunov theory,an existence condition of the desired controller is established first and then the desired controller gains are designed in terms of a feasible solution to a matrix inequality.