From experiment design to closed-loop control (Q1776434)

The paper is a survey of modeling and system identification techniques suited for industrial use and control design applications. General modeling principles and control related issues are overviewed. In particular, such topics as 1) modeling frameworks, 2) model uncertainty, 3) accuracy of approximate models, 4) model validation, 5) links between robust control and system identification, 6) direct identification of a simplified model for control design, 7) approximation of nonlinear systems with linear time-invariant models and 8) model-free tuning methods are discussed in this framework. Moreover, experiment design issues and control oriented model validation are considered for completion. The considerations are in principle restricted to time-domain identification, with particular attention to the prediction error method. The paper concludes with comments on the subject and provides a remarkable number of references.