Awareness and its use in Incremental Data Driven Modelling for Plug and Play Process Control

Abstract

In this paper, we focus on the problem of incremental system identification for the purpose of automatic reconfiguration of control systems. We consider the particular case where a linear time-invariant system is augmented with either an extra sensor or an extra actuator and derive prediction error methods for recursively estimating the additional parameters while retaining the existing system model. Next, we propose a novel measure of the "usefulness" of new signals that appear in an existing control loop due to the addition of a new device, e.g., a sensor. This measure, which we refer to as awareness, indicates if there is a relation between the signal provided by the new device and the existing process, as well as what the new device is good for in terms of control performance. Finally, a simulation example illustrates the potentials of the proposed method.