Multiple-Layer Variable Structure Controller with Parameter Adaptation

Abstract

This paper deals with the design procedure of a multiple-layer variable structure controller (MLVSC) with on-line parameter adaptation. The MLVSC is obtained by non-linear interpolation of the control laws of a set of VSC with sliding mode dynamics, each of them designed on the basis of progressively pessimistic estimates of the parametric uncertainty and disturbance input of the plant. This new structure improves both the robustness and control performance of the single VS controller when poor knowledge on plant parameters and disturbance bounds is available. The control algorithm is completed by an on-line adaptation mechanism aiming to achieve fine tuning of the parameters. This mechanism is based on the trade-off between the ‘chattering’ effect and the tracking error. This heuristics, formulated in terms of ‘if-then’ linguistic rules, is implemented by a fuzzy logic (FL) algorithm. The paper also addresses the stability issue of the closed loop system, giving sufficient conditions for global stability in the presence of bounded disturbances and modelling uncertainties. Finally, a simple simulation example illustrates the main design steps of the control approach proposed.