ABOUT THE COURSE:Data-driven control is the emerging new tool for designing control systems without explicitly constructing the model either from experiments or from first principles. The behavioural approach to systems theory is particularly suited for data-driven control because it puts the “behaviour” of the system, i.e., the set of trajectories that a system admits, at the centre-stage and not its model. In this course we shall first learn the basics of behavioural theory. We shall then use these ideas to explore the data-driven methodologies of control. In the process we shall understand the fundamental idea behind data-driven control, namely, the notion of persistency of excitation.INTENDED AUDIENCE: Anyone with a BE/BTech having done a UG level course on control theory, or with an MSc in mathematics having done courses on dynamical systems can go through this course.PREREQUISITES: Basic course on control theory, preliminary linear algebra.INDUSTRY SUPPORT: Any company that does control systems design: aerospace, robotics, chemical processes, etc. should value/recognize this online course. For example, Eaton, Honeywell, GE, ABB, etc.
Behavioural Theory of Systems with a View Toward Data Driven Control
Indian Institute of Technology Bombay and NPTEL via Swayam
Overview
Syllabus
Week 1: Introduction: Systems, behaviours, dynamical systems. Fundamental properties: linearity,shift/time-invariance.Week 2:Models and representations: AR, MA, ARMA. Kernel, image, latent variables.Week 3:Behaviours and modules: The fundamental principle and elimination theory.Week 4:Two crucial concepts: Controllability and observability.Week 5:From time series to linear systems: the most powerful unfalsified model.Week 6:The fundamental lemma of data-driven approach: persistency of excitation.Week 7:Data-driven control: Data-driven satbilization.Week 8:Data-drive control (contd.): Data informativity.
Taught by
Prof. Debasattam Pal
