Fast-model predictive control of multivariable systems

Abstract

The paper extends a simplified fast-model predictive control technique to a multi-variable system. The strategy is based on identifying drive combinations which will maximise the time at which predicted variables are ‘offside’, having a sign which opposes that of the drive signal. It is relevant to systems such as multi-axis robot configurations, in which each variable can be associated with an input which dominates it, so that, if driven by an input of limiting value, the variable will ultimately take the same sign as the input. Billingsley's SISO strategy is restated and the multivariable strategy, as extended by Mo, is illustrated by a simulation for which results are presented.