State Estimation And Sliding Mode Control Of Complex Uncertain Dynamical Systems

In real control systems,parameter perturbations,disturbances,and even faults,are common in system operation.Moreover,networked control systems(NCSs)are usually characterized by uncertainties such as delay,packet loss,and anonymous attacks.Since these complex uncertainties are likely to affect or even damage the stability and performance of the systems,studies on estimation and control of the dynamic systems with complex uncertainties can promote the development of the system estimation and control methods,and also provide technical support for robustness,fault tolerance,reliability and security of the systems.In this dissertation,the common uncertainties,such as process disturbances,sensor and actuator disturbances and parameter perturbations in physical units,and time delays,packet loss,attacks,quantization,and aperiodic events in network units,are considered,for some typical NCSs.The estimation and control of several dynamic systems with these complex uncertainties are studied,by utilizing mainly the sliding mode control method with auxiliary methods such as adaption,sliding mode observer,and H_?techniques.Some new methods are proposed to solve a series of problems such as stability and performance analysis,state and fault estimations,fault-tolerant control,event-based control,and so on.The specific contents and methods of this dissertation as well as the corresponding theoretical contributions are presented as follows:1.In Chapter 2,a class of uncertain nonlinear systems with parameter perturbation and output disturbance is considered.In terms of the unknown actuator fault deviation,the extended observer technique is used to extend the system state and fault deviation to a new augment state vector,and an extended observer is designed.The observer is a sliding mode observer,which is used to estimate the original system states and fault deviation and compensate for complex uncertainties.By constructing the resulting estimation error system and combining with the observer system,the linear sliding mode surface and the integral sliding mode surface are designed respectively.The integral sliding mode control law and the switching control law are proposed to meet the reachability of the two sliding surfaces.As a result,the purpose of the control of the original system under complex uncertainties is achieved,while the original system state and actuator fault deviation can be estimated simultaneously.2.In Chapter 3,a delta-domain discrete-time system with matching uncertainties and external disturbances are considered.Especially for unknown fault factors of the actuator,each fault factor is described as an unknown constant with known upper and lower bounds.At the same time,the actuator fault deviation is described as an unknown constant with upper and lower bounds,but both are unknown constants.The adaptive laws for estimating the fault factors and unknown upper and lower bounds are designed.By using a linear sliding variable,the conditions for solving the parameters of the sliding surface are derived based on the equivalent control method.Furthermore,an adaptive sliding mode control law based on the linear sliding surface is proposed to realize the active fault-tolerant control for this kind of uncertain systems.3.In Chapter 4,on a class of typical cyber-physical systems(CPSs)with the process disturbance and measurement noise,the integrity attacks(or called the network deception,the false data injection)are analyzed.The discrete-time CPS under the framework of the delta operator is constructed.In terms of the isolated attack of sensor and the joint attack of sensor and executor,a revised delta-domain estimator as well as the one based on matrix enumeration are designed for the secure state estimation of the CPS with integrity attacks.Considering the limited network communication in the CPS,a self-triggering controller is proposed based on the observer.It can not only achieve the almost equivalent control effect and enhance the estimation performance as the time-driven controller but also save some communication resources and improve the utilization of network resources.In the aspect of stability analysis,the sufficient stability conditions of the estimation system are established by using the delta-domain disc stability criterion and the dissipativity theory.The conditions are also derived and then used to solve the parameters of the estimator and the controller.In addition,simplified results of the secure estimation and self-triggering control for the case of the CPSs with isolated actuator attacks are provided.4.In Chapter 5,a class of networked systems with nonlinearities coupled with stochastic processes is considered.In terms of some imperfect logarithmic quantization for the network communication,it is described as a process of random packet loss of quantized signals.Due to the discontinuity of the quantization,a Luenberger observer is constructed to observe the discontinuous quantized values.Then,the sliding mode control and state estimation approaches are proposed based on the state observer.Regarding the stability analysis,the stability of the estimation system and observer system is analyzed by using the Lyapunov stability theory.Meanwhile,the criterion that the output of the system satisfies an H_?performance is established.Besides,a set of solvable matrix inequalities for determining the parameters of the observer and sliding surface are provided.Therefore,the designed observer can estimate the system state,and the designed sliding mode controller can drive the trajectories of the estimated states as well as the system states to the origin.It thus can be used to solve the estimation and control problems of the considered networked stochastic systems under imperfect logarithmic quantization.5.In Chapter 6,single and multiple-loop NCSs with parameter perturbation,external disturbance,and nonlinearities coupled with stochastic processes are further considered.Regarding the aperiodic events between the sensor unit and the controller unit,an observer-based event-triggered sliding mode control is proposed.When analyzing the stability of the single-loop NCS,the considered event-based system is modeled as a stochastic system with state delays by using the idea of“time delay”.According to the method of system stability analysis in Chapter 5,the sufficient conditions of the stability of the estimation system and the original system are obtained by using the Lyapunov-Krasovskii functional method.In addition,the proposed method is extended to the estimation and control of multi-loop NCSs.Considering the shared network communication link from the sensor unit to the controller unit,the multiple access conflict is described as a kind of“queuing delay”,based on the“time delay”concept.A carrier sense multiple access protocol is applied to determine which conflicted data is allowed to send.Then,a multi-loop networked stochastic system with state delays is modeled.Using the analysis method for the single loop networked system,the stability criteria of the corresponding multi-loop networked stochastic system are derived.The corresponding integral sliding mode control law and state observer are also designed.Thus,based on a carrier sense multiple access protocol,a multiloop event-based sliding mode control method is proposed.