An evolutionary based optimisation method for nonlinear iterative learning control systems

Recently, a genetic algorithm based optimisation method for iterative learning control systems (GA-ILC) has been proposed in [5, 6]. The strength of this method is that it can cope to hard constraints in the problem definition whereas most of the existing algorithms would fail. In this paper we extend this method to the case where the dynamical system is nonlinear and it is shown that under suitable assumptions the GA-ILC algorithm will give monotonic convergence. Simulations show that the convergence speed is satisfactory also in practical terms, i.e. it takes less than ten iterations for the algorithm to converge with a nonlinear plant model.