Improved Iterative Learning Control Algorithm And Its Robustness Analysis

Iterative Learning Control?ILC?can effectively deal with the tracking control problem of repetitive controlled systems,and ultimately achieve the tracking objectives set in limited intervals.Most of the other advanced control methods are based on model design,while ILC does not require precise knowledge of the model information of the controlled system.In the process of iteration,only some prior knowledge is needed to achieve the goal of tracking control.When applied to a class of systems with strong non-linear coupling,high position repetition accuracy,high precision trajectory tracking control and difficult modeling,it can show good tracking performance.In the past decades,many scholars at home and abroad mainly optimize the ILC control law to improve the tracking performance of the algorithm,so as to meet the needs of actual industrial processes.This paper mainly studies the iterative learning optimization algorithm,the fast convergence of tracking error can be guaranteed by introducing optimization idea to optimize the control law,and the robustness of the algorithm is discussed.The main work of this paper is as follows:1)To solve the tracking control problem of a class of single-input single-output discrete linear time-invariant systems,an inverse model-based PD-type parameter optimization ILC algorithm is proposed.In the design of the control law,the learning gain matrix is added to the proportion and differential terms of the error,which is obtained by inversion of the established system model.At the same time,the optimal performance index function is constructed to optimize the parameters.so that the tracking error of the next iteration is only related to the parameters to be optimized and the tracking error of the current iteration.In other words,the accuracy of the system modeling does not affect the convergence of the algorithm.Theoretical analysis and simulation results show that the algorithm can be applied to non-positive definite systems,and when there are modeling errors and external disturbances,the algorithm can also show good tracking performance.2)To solve the problem of point-to-point tracking control for a class of discrete linear time invariant systems,this paper combines the point-to-point ILC with the idea of parameter optimization,and proposes a point-to-point high-order parameter optimization ILC algorithm with fast reference trajectory updating.Firstly,when updating the reference trajectory,the fixed learning gain?in the interpolation method is changed to an exponential variable gain with the iteratione^??k?,which makes the new reference trajectory approach the output of the system more quickly,thus reducing the running time of the algorithm and improving the learning efficiency.Then,the input and output information obtained in multiple iterations are used to construct the new control input,and the idea of parameter optimization is introduced to ensure optimal operating conditions.Finally,in order to verify the effectiveness of the algorithm,the tracking control of the motor-driven single manipulator control system is used to verify the performance of the algorithm.3)For a class of general single-input single-output discrete-time nonlinear systems,this paper introduces the idea of optimization and data-driven control into ILC,and proposes a data-driven parameter optimization adaptive iterative learning control algorithm.The algorithm adopts the traditional PD iterative learning control law,and introduces the idea of parameter optimization and data-driven control to improve the tracking performance of the algorithm.Some parameters of the system need to be used in the process of parameter optimization,but it is difficult to model the non-linear system.To solve these problems,data-driven control is introduced to estimate the parameters of the system.With the iteration,the system parameters can be accurately estimated to ensure that the parameter optimization theory can also be applied in non-linear systems which are difficult to model.The whole process is based on the known input and output information,so the algorithm is data-driven and has low dependence on the accuracy of system modeling.Theoretical analysis and simulation results show that the algorithm can show good tracking performance for a class of nonlinear systems which are difficult to model or have modeling errors,even if there are input and output disturbances in the tracking control process.