Iterative Learning Control Laws Design Based On Optimization Based On Optimization

The choice of parameters utilized in the Iterative Learning Control (ILC) law has a great influence on the algorithm convergence and its convergence rate. But in classical PID-ILC, parameters are set under experience, it will cause some blindness. In order to overcome the blindness of parameters setting, an effective method is to design the optimal ILC Law based on system model knowledge and optimization criterion, which can ensure automatic choice of step size and improve the convergence rate by simple way. Therefore, it is meaningful to study the ILC theory based on optimization criterion.After deeply studying the status of optimal ILC, in this paper, the methodology of constructing iterative learning control law based on optimization is researched for the linear and nonlinear system respectively. The main results are as follows:1. Under the ILC structure combined feedforward with feedback control, an optimal ILC laws with state observer is designed based on optimization criterion for continuous linear system, and the algorithm convergence condition is obtained, then the simulation in injection ram velocity control is given.2. An optimal ILC based on feedforward and feedback control structure is proposed for discrete linear system, and two formulation forms to compute the feedforward action value is given, and a method for introducing state observer to construct the optimal iterative learning law is illustrated, then the convergence condition and simulation results are shown.3. For a kind of nonlinear discrete system, two training methods of control optimization and parameters optimization are proposed based on optimization criterion, respectively, and the problem with hard input limit is solved through constructing constraint condition in optimization. Meanwhile, the algorithms application effect is validated through simulation.