Research theme: System identification

When an actual plant exists, the problem of developing a model for a model based controller is addressed primarily based on the signals acquired from the plant to be controlled. In the development of solutions for such identification problems research is conducted along 3 different approaches: First, the intelligent modeling approach. Intelligent methodologies employ techniques inspired by the functionality of intelligent biological systems. Second, to improve the user interaction and performance of identification methods novel techniques for specific classes of linear, non-linear and hybrid systems are developed. Third, the key question in identification for control is the development of low order models based on the design of appropriate experimental conditions and closed-loop system identification techniques. Considering applications with arrays of thousands and thousands of sensors (MEMS) and actuators the analysis paradigm is shifted from centralized processing methods towards decentralised ones.

• Model based monitoring of large scale processes in process industry
• Towards experiment-based robust control design
• Identification of hybrid systems
• Bounding uncertainty in subspace identification
• Identification of nonlinear state-space systems
• Development of computationally efficient and numerically robust system identification software
• Choice of uncertainty structure in identification for robust control
• Multi-objective nonlinear identification
• System identification of bio-technological processes
• Classification of buried objects based on ground penetrating radar signals
• Identification and control of LPV systems using orthonormal basis functions