Networked and Distributed Control Systems

SC42100



ECTS:

3

Responsible Instructor:

dr.ir. T. Keviczky (Tamas)

Responsible for assignments:

Prof.dr.ir. N. van de Wouw

Contact Hours / Week x/x/x/x:

0/0/0/4

Education Period:

4

Start Education:

4

Exam Period:

4

Course Language:

English

Course Contents:

This course starts with the modelling of so-called networked control systems (NCS). Networked control systems are systems in which the communication between plant and controller takes place via a (e.g. wireless) network. Such network-based communication leads to imperfections in sensor and control signals, such as time-varying and uncertain sampling intervals and delays, packet dropouts, scheduling constraints, etc. The modelling framework introduced in the course is subsequently used to support stability analysis, using characterizations based on linear matrix inequalities (LMI). Such stability analysis allows to study trade-offs between requirements on the controller, the network and plant properties.

The second part of the course deals with the aspect of the distributed control of networked systems. In particular, distributed optimization methods and various decomposition techniques (primal, dual, augmented Lagrangian / proximal point method, ADMM), links to consensus algorithms, and their application in networked multi-vehicle distributed robotics problems. Online optimization-based control approaches such as distributed model predictive control for multivehicle cooperation, distributed LQR and decomposition based methods that are applicable to collections of mobile agents. The methods will be illustrated on application examples including cooperative rendezvous, distributed formation control, spacecraft formation flight, and robotic networks.

Study Goals:

The student must be able to:
1. model networked control systems with network-induced uncertainties / effects / imperfections, including time-varying and uncertain sampling intervals, delays, packet dropouts, and scheduling constraints
2. analyse the stability of NCS (involving the above effects), e.g. by applying LMI-based stability characterizations
3. describe and apply decomposition techniques for distributed optimization to various examples
4. describe and apply consensus algorithms to multi-agent coordination problems
5. solve cooperative control problems by implementing a distributed model predictive control approach
6. analyse the stability and convergence of distributed control methods that rely on online optimization

Education Method:

Lectures

Computer Use:

Matlab/Simulink is used to carry out the exercises of this course.

Literature and Study Materials:

Course material:
Lecture slides including additional reading material in the form of papers and textbooks are made available online.

Assessment:

Project (in the form of several homework sets) + Oral Exam

Department:

3mE Department Delft Center for Systems and Control
© Copyright Delft Center for Systems and Control, Delft University of Technology, 2017.