Real-time hybrid simulation of an adaptive-control technique

Abstract

The paper describes the simulation of an adaptive scheme based on least-squares methods, incorporating a hybrid computer. The analogue computer simulates the process dynamics, while the online digital machine effects the adaptive control via a.d.c. and d.a.c. interfaces. The scheme is shown to be realistic, and to have application to relatively fast-responding systems. The process is repeatedly identified as a series of quasilinear systems. Derived process parameters continuously update a suboptimal-feedback-control law. This control law, valid for a discrete-time environment, is shown to have application in continuous time. In addition, a novel approach to the problem of inaccessible states is investigated.