Direct predictive adaptive control with stochastic gradient identifiers

Abstract

The multistep multivariable adaptive regulator, which is a direct predictive adaptive controller, is considered. Specifically, the replacement of recursive least squares (RLS) identifiers with stochastic gradient (SG) identifiers in that controller is considered. By an ordinary differential equation analysis, local convergence properties of the algorithm are investigated. The conclusions, supported by various simulation results, are that, in contrast to the case of the one-step-ahead self-tuning regulator, replacement of RLS with SG deteriorates the convergence properties of the resulting adaptive controller.<>