Direct adaptive multiple-model control schemes

Abstract

In this paper a direct model reference adaptive control based multiple-model control framework is developed. Such multiple-model control scheme is designed for single-input single-output (SISO) systems, and its basic design procedure and new features are demonstrated. Such a scheme is then developed for multi-input multi-output (MIMO) systems, showing its key advantage of capable of ensuring the nonsingularity of a gain matrix estimate, crucial for multivariable adaptive control, over other indirect adaptive multiple-model control designs. In these schemes, a bank of adaptive controllers initialized from different parameter subregions are constructed whose parameters are directly updated, not indirectly calculated from plant or certain reparametrized system parameter estimates. A control switching mechanism is designed with performance indices based on tracking error related estimation errors. An input filtering method for smooth control is proposed for those applications whose input signals need to be continuous. Simulation results for a SISO system are presented to show the improved system transient responses.