PhD course in diagnosis, 9-12 credits,
Vehicular systems, VT2010

General information

A PhD course in diagnosis is given by Vehicular systems, Department of Electrical Engineering at Linköpings universitet during spring 2010. The number of credits all depend on degree of participation, typically 9 to 12 credits. Of course there are possibilities to, after discussion with the course leaders, obtain more credits if you choose to extend the course with your own projects.

A pre-requisite is the undergraduate course "TSFS06, Diagnos och övervakning" or corresponding.

Course leaders are Erik Frisk and Mattias Krysander.

Course objectives

The goal of the course is to give deepened and more general knowledge in a number of areas all fitting under the headline diagnosis. This is therefore not a progression based course like the undergraduate course, but is more of a methodology course that complements the knowledge obtained in the basic course. See the brief outline of the course below, which includes the course contents, for a list of topics covered by the course.

Course organization

The course is organized with no coventional lectures, if needed presentations by the course leaders on specific topics can be arranged. Instead, the course is based around discussion seminars and exercise seminars.

Planned start is mid February 2010.

Examination

1. Active participation in discussion and exercise seminars. Attendence is obligatory (for some reasonable definition on obligatory).
2. Act as discussion leader for a number of seminars (how many depends on the number of course participants and will be decided upon course start). This includes preparing a discussion material to be handed out to all course patricipants in advance. Contents of the discussion material will be specified by the course leaders at the start of the course.
3. Handing in a report with solutions for examination assignments together with written code. There will be relatively few examination exercises, but slightly larger than regular exercises and typically includes implementation of some method.

Course material

The course material will consist of papers and books that will be specified during the course. Links to PDF-files with the material will be found at a course material page. See course material page for material used last time.

Brief outline of the course

The course includes 6 parts, where the purpose of the 2 last ones is to give a brief introduction. In addition to the course outlined below, a part with probabilistic fault diagnosis based on Anna Perneståls Phd thesis might be given also.

• Principles of model based diagnosis
  • Consistency and abductive based diagnosis.
  • Definitions and theory about diagnosis, minimal diagnosis, partial diagnosis, and kernel diagnosis.
• Fault isolation algorithms
  • Algorithms for computing multiple fault diagnosis given conflicts.
  • Fault isolation algorithm handling fault modes.
  • Focusing strategies.
• Residual generation techniques
  • Optimal linear residual generation. Frequency and time domain approaches and their relation.
  • Nonlinear residual generation, analytical redundancy relation and observer based approaches.
• Analysis methods
  • Two graph representations of structured systems.
  • Different application of structural analysis: Computing overdetermined sub-systems, fault detectability and isolability analysis, and sensor placement analysis.
• Selected diagnosis tasks.
  • Troubleshooting.
  • Fault tolerant control.
• Model-based diagnosis for different type of systems.
  • Discrete-Event Dynamic-Systems (DEDS)
  • Hybrid systems/TRANSCEND/Bond graphs