An optimization-based anti-windup design for MIMO control systems

This paper presents an optimization-based anti-windup design generalized from existing conditioning techniques to provide almost equally good performance over a large operating range. Hanus' conditioning technique is known as an effective anti-windup design but is liable to shortsightedness. Walgama's GCT is an effective solution to the shortsightedness problem for a limited operation range, but not sufficiently effective in a large operating range. To overcome these shortages, two analyses of shortsightedness are presented in this paper to indicate the causes of shortsightedness in the existing conditioning technique-based approaches. First, a normalized impulse response analysis reveals that a suitable factorization of the controller is helpful but not sufficiently effective in alleviating shortsightedness. Next, a phase plane analysis shows that shortsightedness is mainly due to the use of an improper objective function in an optimization problem which determines a realizable control input. Based on the previous analyses, this paper presents a generalized optimization-based conditioning technique that is capable of providing almost equally good performance over an excessively large operating range.