State,Input And Fault Estimation With Communication Constraints

In real engineering systems,the external input and the fault are nonnegligible factors that affect the operation of the concerned systems.In many scenes,the information of inputs and faults is unknown and it is usually difficult to precisely obtain the information.To further analyze and control the systems,it is inevitable to estimate the states,unknown inputs,and faults simultaneously.On the other hand,the network communication technique has been widely applied in the fields of traditional control and filtering.As a result,most of the data(including control signals,measurement outputs)are transmitted via the communication network.Due mainly to the constraints on the communication environment and equipment,the communication resources are essentially limited,which leads to the communication constraints in data transmissions,including the data dropouts,the limited transmission energy,and so on.Undoubtedly,the communication constraint phenomenon challenges the traditional control and filtering theory.Therefore,it is of great significance in both application prospect and theoretical value to develop the estimation methods of state,unknown inputs and faults under communication constraints.In this thesis,the states,unknown inputs and faults estimation problems are investigated for stochastic systems under communication constraints.The content of this thesis can be mainly divided into three parts.In the first part,we consider the event-based state and input estimation problem with communication constraints,and both the effects of the static event-triggering mechanism and the dynamic eventtriggering mechanism are analyzed simultaneously.In the second part,the unbiased optimal state and unknown input estimation problem is investigated for stochastic systems with random data dropouts,and the distributed estimation problem of the state and input is further addressed for the sensor networks.In the third part,we focus on a kind of special unknown input,i.e.,the faults,and study the H_∞fault estimation problem for uncertain systems with missing measurements.The compendious frame and description of the thesis are given as follows.· The event-based state and unknown input estimation problem is investigated for a class of stochastic systems subject to parameter uncertainties and stochastic nonlinearities.For the purpose of reducing the energy consumption in data transmission,an event-triggering protocol is employed to regulate whether the current measurement is transmitted by the sensor.Utilizing the event-triggered measurement,a recursive estimator is constructed to concurrently estimate the state and the unknown input.The upper bounds of estimation error covariances are given explicitly for both the state and the unknown input estimates.By means of the computing-the-square technique and Lagrange multiplier method,the estimator gain matrices are designed which minimize the obtained upper bounds.· The dynamic event-triggered state and input estimation problem is considered for nonlinear systems with random packet dropouts.The dropouts are essentially characterized by random variables following the Bernoulli distribution.Moreover,a dynamic event-triggering mechanism is adopted to determine when the current measurement is spread for the sake of saving the energy of sensor nodes.By utilizing the mathematical induction method,upper bounds are obtained for state and unknown input estimation error covariances.Then,the desired gain matrices are constructed by minimizing the derived upper bounds.· The state and unknown input estimation problem is investigated for a class of stochastic systems with energy-harvesting techniques.An augmented system is introduced to analyze the dynamics of the unknown input and states simultaneously.Moreover,a random variable,which obeys the Bernoulli distribution,is employed to characterize whether the signals are transmitted successfully.Then,the probability distribution function of such a random variable is derived.By utilizing the Kalman filtering strategy,an upper bound is obtained for the filtering error covariance and the gain matrices are further designed to minimize the obtained upper bound.· The simultaneous input and state estimation problem is studied for a class of linear discrete-time systems with missing measurements and correlated noises.The missing measurements occur in a random way and are governed by a series of mutually independent random variables obeying a certain Bernoulli distribution.The process and measurement noises under consideration are correlated at same time instant.Our attention is focused on the design of recursive estimators for both input and state such that,for all missing measurements and correlated noises,the estimators are unbiased and the estimation error covariances are minimized.· The distributed unknown input and state estimation problem is addressed for a class of linear discrete time-varying systems in sensor networks with missing measurements.Sensor nodes under consideration are distributed in space according to a fixed network topology,and each sensor receives the information from both the system and its neighbors.The missing measurement phenomenon on each sensor occurs in a random way and is governed by a series of mutually independent random variables obeying a certain Bernoulli distribution.Our attention is focused on the design of a recursive estimator for every sensor node to simultaneously estimate unknown input and state in the sense of unbiased minimum-variance.Using direct algebraic operation,we obtain the estimator parameter matrices recursively,and the design algorithm is also provided.· The H_∞ fault estimation problem is investigated for a class of uncertain linear discrete time-varying systems with missing measurements over a finite timehorizon.In order to characterize the random nature resulting from the missing measurements,the conventional H_∞performance index is first modified in the stochastic sense.Then,by introducing an auxiliary system,the parameter uncertainties are eliminated formally and the modified H_∞performance index is described by a certain indefinite quadratic form.Subsequently,the desired fault estimator is designed by equivalently solving a minimum problem of the indefinite quadratic form.