DOI: 10.5937/fme2604588f ISSN: 1451-2092
Adaptive one-step-ahead iterative learning control based on generalized forgetting recursive least squares
Vojislav Filipović, Milan Matijević, Dragan Kostić, Saša Ćuković, Mihailo LazarevićThis paper considers adaptive one-step-ahead Iterative Learning Control (ILC) for systems described by a Deterministic Autoregressive Moving Average (DARMA) model. The unknown model parameters are estimated using the recently introduced generalized forgetting recursive least squares method. The system is assumed to be minimum phase. The paper relaxes two fundamental ILC assumptions: identical initial conditions and identical reference trajectories. Instead, the initial regressor vector is considered random, and the reference trajectory varies across iterations. The proposed algorithm has a 2D structure, where parameter updates are performed iteratively along the iteration axis and recursively along the time axis.