Modeling and Estimation

at the Department of Control Engineering, Aalborg University

Introduction

In engineering practice, a major part in the process of designing a control system involves deriving a nominal model of and (possibly) an uncertainty model for the process. Therefore it is important for a department in control engineering to accumulate expertise in this area. Moreover, in addition to controlling a dynamical process, it is often required to estimate states of which no or only very noisy measurements are available.

Modeling and estimation are independent research areas, interesting in their own right. However, in the present research strategy these areas are considered subordinate to control engineering.

This research area covers the following three subtopics:

Grey box modeling
System identification
State estimation

Classically, two major methodologies have been developed for deriving models of a dynamical process: first principles modeling and system identification. In first principles modeling procedures, the model is derived from basic physical laws only. However, for complex industrial processes, it is rarely possible to derive a complete model of the process by first principles modeling. Therefore, in practice first principles modeling is combined with system identification methods. The theoretical framework for doing this systematically is known as grey box modeling.

The scope of System identification is to derive mathematical models mainly based on experimental data. If nothing at all is known or assumed on the physical structure of the system modeled, the modeling procedure is known as black box modeling.

In addition to the role played by models in feedback control, models are also used in industry for estimation purposes. Estimators have proven to be extremely useful in a wide range of applications, including noise reduction of signals, sensor fusion, trajectory tracking, and control of linear and nonlinear systems.

The Department of Control Engineering has been very active in adaptive identification methods. These are described in the control theory part of the strategy.

separator

Projects

Significant elements of modeling and estimation are found in practically all projects at the Department of Control Engineering, whereas few projects are dedicated entirely to the study of modeling and/or estimation.

Relevant projects are: