Robust Design of Adaptive Control Systems Using Conic Sector Theory

Abstract This paper presents a design approach for discrete adaptive control systems which provides a quantitative measure of the effect of design alternatives such as (i) adaptive gain, (ii) model order, and (iii) sampling rate, on stability in the presence of unmodeled plant dynamics. The proposed method, based on the conic conditions developed by Ortega and colleagues (1985), is illustrated using Rohrs' (1982) benchmark example. The results demonstrate that the sector conditions permit design tradeoffs to be made such that stability is maintained despite the model-plant mismatch.