Identification and Control of Systems With and Without Zeros Via Approximation of the State Evolution Function

Abstract

Solution to the identification and control problems for linear and non linear systems is found using state-space theory and tools of artificial intelligence. We review previous results using the Hybrid State Transition Kernel, which has as distinctive features 1) requires little a priori information, 2) is independent of the approximation or learning method, 3) uses the same data for identification and control, 4) it implements inverse plant control with finite convergence time. The structures necessary to treat plants with and without zeroes are derived, showing simulation results for linear and nonlinear systems.