Parameter Optimal Iterative Control

Abstract

NOILC is, typically, based on function or time series optimization. An alternative parameter optimization approach is derived based on an equivalence theorem for finite dimensional (discrete) systems. This is used to construct simple Iterative Algorithms with monotonic convergence properties. Emphasis is placed on single parameter controllers although the ideas apply equally to multi-parameter cases. Convergence properties are established and the possibility of applications exhibiting flat-lining phenomena is discussed. The use of switching algorithms is shown to offer a means of ameliorating this behaviour and recovering the desired error convergence. Algorithm robustness is analyzed using links to the fixed parameter studies in previous chapters.