Courses in previous years: [ 2000 | 2001 | 2002 | 2003 | 2004 ]

T-61.6050 Special Course in Computer and Information Science V L:

Neural Networks for Modelling and Control of Dynamic Systems 5 cr

Notices (updated 19.10.2005) * We will start the lectures strictly at 14.00. The change is valid for the whole semester. * Information and description of the project is added on the web page (see. section Project work). * The project work has been changed. You can find the description of the project from the section Project work.

Lecturers	Prof. (pro tem) Jaakko Hollmén, PhD (Eng.) Amaury Lendasse
Assistant	M.Sc. Jarkko Tikka
Credits (ECTS)	5
Semester	Autumn 2005 (during periods I and II)
Seminar sessions	On Wednesdays at 14-16 in lecture hall T4 in computer science building, Konemiehentie 2, Otaniemi, Espoo. The first session on 14.9.2005.
Language	English
Web	http://www.cis.hut.fi/Opinnot/T-61.6050/
E-mail	tikka (at) mail.cis.hut.fi

Introduction

The technology of neural networks has attracted much attention in recent years. Their ability to learn nonlinear relationships is widely appreciated and is utilized in many different types of applications; modelling of dynamic systems, signal processing, and control system design being some of the most common. The theory of neural computing has matured considerably over the last decade and many problems of neural network design, training and evaluation have been resolved. This course provides a comprehensive introduction to the most popular class of neural network, the multilayer perceptron, and shows how it can be used for system identification and control. It aims to provide the student with a sufficient theoretical background to understand the characteristics of different methods, to be aware of the pit-falls and to make proper decisions in all situations. The subjects treated include: System identification: multilayer perceptrons; how to conduct informative experiments; model structure selection; training methods; model validation; pruning algorithms. Control: direct inverse, internal model, feedforward, optimal and predictive control; feedback linearization and instantaneous-linearization-based controllers. Case studies: prediction of sunspot activity; modelling of a hydraulic actuator; control of a pneumatic servomechanism; water-level control in a conical tank.

Table of Contents

1. Introduction
2. System Identification with Neural Networks
3. Control with Neural Networks

Course format

Seminar course

Requirements for passing the course

Detailed requirements are given in the first lecture.

Course material

Neural Networks for Modelling and Control of Dynamic Systems
Magnus Nørgaard, Ole Ravn, Niels K. Poulsen and Lars K. Hansen
Springer-Verlag, London, 2000
(68 EUR in AKATEEMINEN KIRJAKAUPPA)

Schedule

Time	Lecturer	Subject
14.9.	Amaury Lendasse	Presentation of the course
21.9.	Elia Liitiäinen Esa Seuranen	Introduction to ANN and System Identification p.1-18 + paper 1, Esa + Elia
28.9.	Vibhor Kumar Sven Laur	Model Structure Selection p.18-37 + paper 1, Vibhor + Sven
5.10	Yoan Miché Kei Takahashi	Experiments and Determination of Weights p.38-84, Yoan + Kei
12.10	Rami Rautkorpi Ali Pekcan	Validation Procedure and Summary p.85-119, Rami + Ali
19.10	-	Project
26.10	-	Project (no lecture)
2.11	-	Project
9.11	Mikael Pohjola Janne Pylkkönen	Introduction + Direct Inverse Control + IMC p.121-142, Mikael+Janne
16.11	Antti Yli-Krekola Eemeli Aro	Feedback Linearization + Feedforward Control + Optimal Control + CBIL p. 143-175, Antti+Eemeli
23.11	Antti Sorjamaa Ji Yongnan Nima Reyhani	Predictive Control + Recapitulation + Case Study p. 178-233
30.11		Presentations of projects (20 minutes per group)

paper 1:
Nonlinear Black-Box Modeling in System Identification: A Unified Overview
Jonas Sjöberg, Qinghua Zhang, Lennart Ljung, Albert Benveniste, Bernard Delyon, Pierre-Yves Glorennec, Håkan Hjalmarsson and Anatoli Juditsky
Automatica, Volume 31, Issue 12, Pages 1691-1724, Elsevier, December 1995

Project work

The project by group of 2 or 3 students during weeks 6 to 8 IN THE CLASSROOM.

Identification and control of a simulated dynamical system.
Tool: NNSYSID and NNCTRL Matlab toolboxes

Information and description of project work: project.pdf or project.ppt
Data for the project: dryer2.mat
Deadline for the project is November 9th at 2 pm.

For more information, please send email to the course assistant (tikka (at) mail.cis.hut.fi).