COURSE SYLLABUS

Estimation in Control 7.5 Credits

Estimering i reglersystem
Second cycle, R7011E
Version
Course syllabus valid: Spring 2017 Sp 3 - Present
The version indicates the term and period for which this course syllabus is valid. The most recent version of the course syllabus is shown first.

 Education level Second cycle Grade scale G U 3 4 5 Subject Control Engineering Subject group (SCB) Automation Technology

Entry requirements

* (prerequisite for the course R7003E) intermediate level knowledge in the subject of Automatic control, specifically regarding frequency response, state-space form, and state feedback; * (prerequisite for the course R7003E) experience with using Matlab for analysis of control systems; * documented skills in English language; * (suggested) basic level knowledge about Probability and Statistics.

Selection

The selection is based on 20-285 credits

Course Aim
The student should be able to:
• formulate and implement algorithms for system identification, i.e. estimation of mathematical models of a dynamic system from input-output data
• formulate and implement algorithms for state estimation, i.e. to infer the status of the internal variables of a dynamic system using measurements of other quantities and some knowledge of the system dynamics
• solve simple instances of system identification and estimation problems by hand
• analyze and prove properties of system identification and estimation algorithms
• apply the above described techniques on real-world processes, and report on this work both orally and in writing.

Contents
The course covers the essentials of two interconnected topics: system identification and state estimation.

System identification is the science dealing with how to model systems starting from collected evidence. Among the statistical sciences, this branch is the one most related to automatic control. Indeed developing a control system usually starts with a system identification step: there is a process to be controlled, but there is either no model for it, or an incomplete model where some parameters are unknown, or maybe there is a model, but it is too complicated for developing a controller (for example a finite element simulator of the thermal dynamics of a whole datacenter, and you want to control the temperature of the racks).

State estimation is instead dealing with reconstructing information on the state of a system starting usually from indirect measurements. For example, gyroscopes are usually subject to some bias, but this bias cannot be measured directly. It is nonetheless possible to infer it indirectly combining knowledge of the dynamics of the system and measurements from the sensors. This state information is then useful for performing automatic control tasks, e.g., feedback from the state.

Realization
The teaching consists of lectures and problem seminars. Lab work and project assignments are performed in groups of no more than two students and accounted for with written reports and a demonstration.

Examination
Written exam with differentiated grades and approved lab work.

Examiner
Damiano Varagnolo

Literature. Valid from Spring 2017 Sp 3 (May change until 10 weeks before course start)
Literature:
- L. Ljung, System Identification: Theory for the User, 1999, Prentice Hall;
- lecture notes that will be made electronically available to the students.

Course offered by
Department of Computer Science, Electrical and Space Engineering

Items/credits