The Application Research Of Iterative Learning Control In Non-strict Repetitive Time-delay Systems

The robustness,adaptiveness and convergent performance of iterative learning control system are analyzed based on non-strict repetitive systems.Simulation and experimental results are shown to valid the conclusions in this thesis.The repeatability of system should be guaranteed for the convergence analysis,which is also a key factor for system performance.However,in reality,because of time-delay,uncertainties and other complex factors as well as the heredity of the non-Markovian system(especially for fractional-order systems),it is hard to guarantee the repeatability of system which may lead to poor performance.Attempt to guarantee the repeatability of the system is even harder than control system design.The system performance in both theory and application is hard to be satisfied for the traditional iterative learning control for non-strict repetitive systems.Although the feedback method and various strategies in robust and adaptive area can be applied to improve the performance of iterative learning control system,this is not a solution in essence.How to decrease the influence of non-repeatability in iterative learning control processes is of great importance,it is also a significant complement for the current learning control scheme.Time-delay is a key factor for the non-repeatability so that the time-delay sys-tem is a typical non-strict repetitive system.In this thesis,the time-constant delay system and time-varying delay system are analyzed,then a general non-strict repetitive system is investigated.Several important iterative learning con-trol approaches are designed and applied to time-delay systems,including linear iterative learning control,nonlinear parametric iterative learning control and ro-bust and adaptive iterative learning control.As a feedforward control method,the robustness of iterative learning control needs to be improved.The design of robust and adaptive iterative learning control is an effect way to improve the performance of the non-strict repetitive system.The output reference iterative learning control is proposed to satisfy the exact tracking process as the tracking trajectory is unknown but guarantees learning laws.The convergence condition and convergence speed are included in the conver-gence analysis of this thesis.Although the exact model of the system is not necessarily required for iterative learning control,allow for various uncertainties and complexity of the system,the system structure,some key parameters and control gains are helpful to improve the performance of the entire control system.To do so,an estimation strategy is proposed for the identification of some key parameters and to design the controller which can further improve the system performance.The preconditioning of non-strict repetitive system(integer-order and fraction-order system)is analyzed.The definition of initialized systems and initialization are shown to illustrate how the history before the initial time in-stant influences the current control processes.The preconditioning process can improve the performance of iterative learning control system.In summary,some urgent problems about non-strict repetitive systems in iter-ative learning control field are analyzed in theory,simulations and experiments.The time-delay systems which belong to a typical non-strict repetitive system and a generalized non-strict integer-order and fractional-order systems are analyzed.The convergence band is studied to expand convergence domain;the applica-tion of feedback control strategy enhances the robustness and adaptiveness;the identification of key parameter improves the convergence speed;the control per-formance is improved to a large extent by initialization and preconditioning which build a bridge between the theory and application which is of great significance.The conclusions are valid by simulations and semi-physical simulations for ma-nipulator model;some experiments for NAO robot and grinding are shown to demonstrate the theory conclusions of non-strict repetitive systems with itera-tive learning control which is of great importance for the application of iterative learning control.