Gain bounds in the minimal controller synthesis algorithm

Abstract

Owing to its high responsiveness to noise disturbances, the minimal controller synthesis (MCS) algorithm can occasionally suffer from the long-term effect of measurement errors, resulting in gain wind-up, closed-loop signal aliasing and hence instability. To counteract this effect, an algorithm is proposed that extends MCS in such a way as to confine the adaptive gain evolutions by using automatically adjusting bounding ‘corridors’. The proof of stability is based upon the original version for MCS, and an implementation study supports the theoretical assertions that the modified algorithm prevents gain wind-up without significantly compromising adaptive effort.