Iterative learning control for linear continuous-time switched systems with observation noise

The paper attempts to explore the effect of an internal switching mechanism and external noise on the learning performance of a conventional proportional-derivative iterative learning control scheme for linear continuous-time switched systems. The learning behaviour is analysed by virtue of a Lebesgue-p norm and the generalized Young inequality of the convolution integral. The results convey that both asymptotic convergence and robustness of the control algorithm depend mainly on the learning gains and the dynamics of the subsystems, but less on the time-driven switching rule. Given an arbitrary time-varying switching rule, a group of learning gains may be properly chosen such that both convergence and robustness of the control algorithm are guaranteed. Numerical simulations verify the validity and the effectiveness.