Event-triggered And Dynamically Quantized Control Of Uncertain Switched Systems

With the engineering and industrial systems becoming large-scale,networked,intelligent and the requirements/environments of which getting more complicated,the networked control problems of switched systems are turning into international frontiers and challenging hotspots.On one hand,switched systems can be used to model and describe a family of continuous dy-namics and the discrete switching events among them.It has been proven that the overall per-formance of a switched system is no longer the composition of its subsystems.When introduced as a control method,switching behavior also enriches the design of system control.On the other hand,one main characteristic in networked control is the resource sharing which includes the calculation and transmission bandwidth.An appropriate sampling mechanism and an efficient quantization strategy are the foundation to ensure the overall performance of networked control systems by using the limited resources.In the existing literature,the research on the problems of event-triggered control and finite-level dynamically quantized control of switched systems is still in its infancy and has many restricts.A number of issues induced by the coupling of switch-ing,event-triggering and dynamic quantization need to be addressed urgently.This dissertation aims to solve the aforementioned problems and the work is summarized as follows.We consider the event-triggered and dynamically quantized control design for a class of switched linear systems whose full state is measurable.The switch inside the sampling inter-val can lead to mismatch between the model and its corresponding controller.This behavior is formulated and analyzed.By designing a triggering mechanism depending on both time and trig-gering error,the mismatch-induced effects to the system performance are blended into the design of update law of quantization parameter.Then,we co-design the quantization policy,triggering condition and dwell-time switching law such that the switched system is exponentially stabi-lized.Based on this study,the sliding-mode control of switched linear systems with unknown disturbances is further considered.By developing a novel level-set method for both sliding vari-able and state component,a dynamic quantization algorithm is designed to address the difficulty of avoiding quantizer saturation in sliding-mode control.The designed new time-varying up-date coefficient and switching algorithm can reduce the design conservatism,compared with the constant one,and the dwell-time constraint of switching signal.For the disturbed switched linear systems whose state is unmeasurable,a robustly switching current estimator is proposed.According to the current error of estimated output,we design a switching estimator input by which an effective scheme is provided to address the state estima-tion of disturbed switched systems.Under different switching constraints,sufficient conditions are given to guarantee the exponential stability of error systems.The norm boundedness for dis-turbances is not needed.For slow-varying disturbances,we further study the quantized output feedback control based on current estimator.An event-driven update policy of dynamic quanti-zation is developed to conquer the challenge,affected by switches and disturbances,in avoiding the saturation of output quantizer.Compared with the existing literature on output quantiza-tion,our method relaxes the structure constraint for the network and removes the requirement of detecting quantizer saturation on-line.The system is eventually stabilized under the designed switching law.Based on the above results on the networked control of switched linear systems,we further consider the event-triggered control for switched nonlinear systems with parametric uncertain-ties.Using the deviation of sub-controllers at switching instant,we construct a novel switching event-triggered mechanism.By co-designing the adaptive control law and triggering parameters,a positive minimum inter-trigger time is obtained and the tracking error of system can be regu-lated to the neighborhood of the origin.This design scheme can address the mismatch problem between nonlinear models and nonlinear controllers in the network.The assumption/restriction for the maximum inter-trigger time is removed.At last,the obtained result is applied to the simulation of tracking problem of a single-link manipulator.To fulfill the demand of certain transient performances,e.g.,settling time,for real-word systems,we consider the finite-time stability of a class of switched nonlinear systems with state jumps.By the dwell-time technique and mathematical induction,the finite-time stability criteria are established for switched nonlinear systems under different types of subsystems.The settling-time formulas for system state converging to the origin are provided.The obtained results can relax the assumption for system structure and the cyclical constraint of switching signal.The simulation verification adopts the finite-time stabilization of a continuous stirred tank reactor and the finite-time consensus of a multi-agent system.