Reuse of constraint knowledge bases and problem solvers explored in engineering design

Peter M.D. Gray, Trevor Runcie, Derek Sleeman
  • Artificial intelligence for engineering design analysis and manufacturing, April 2014, Cambridge University Press
  • DOI: 10.1017/s0890060414000134

Automating the reuse of constraint knowledge by different problem solvers

What is it about?

We have taken knowledge about a classic AI configuration problem (VT - parametric design of elevator) and automatically converted it for use by two solvers - a spreadsheet (Excel) and a Constraint logic solver (ECLiPSe). The method works for a wide range of numerical and logical constraints which in this task represent physical entities as ontological attributes and relationship stored as part of a Protege Knowledge Base. .

Why is it important?

Advances in Constraint Logic Programming have not been widely taken up in engineering design, probably because the declarative programming style is unfamiliar to C and Fortran users. This method generates the constraint program, producing checkable code which can be executed directly. The code can easily be embedded in other systems such as Branch and Bound, or an interactive display. This is a big advance on a formal mathematical notation that is very hard to convert for a solver. It also makes available a constraint propogation technique developed by Le Prevost and Wallace that produced an enormous speedup in solving, from many hours down to seconds. This makes very good use of information stored as tables of parameter values for alternative components, as is widely used in design problems.


Peter Gray
University of Aberdeen

Making really good practical reuse of design data stored in large knowledge bases has been a long term AI research goal of two of the authors (Sleeman and Gray) and it has been very good to work with our third author (Runcie) to make this happen, and to make use of his practical engineering background.

Read Publication

The following have contributed to this page: Peter Gray