Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.639229
Title: Examination scheduling using simulated annealing
Author: Thompson, J. M.
Awarding Body: University College of Swansea
Current Institution: Swansea University
Date of Award: 1995
Availability of Full Text:
Access through EThOS:
Abstract:
Recent increases, both in student numbers and choice of subjects, have made the problem of producing university examination timetables increasingly difficult. Additionally, there is pressure to give students free periods between exams to aid their revision. Many other constraints may also be imposed on the timetable. The aim of this research is to investigate the usefulness of simulated annealing, as a basis of an efficient solution method, for this problem. The examination timetabling problem is shown to be equivalent to the graph colouring problem, with side constraints. This model allows several of the constraints to be included directly, thereby simplifying the problem. The colouring problem is solved in two phases, with the first obtaining a feasible colouring and the second optimising further objectives while maintaining feasibility. Simulated annealing is used to solve each phase and computational experience demonstrates that this results in high quality solutions. In phase two, only feasible solutions, that is those which satisfy the binding constraints, are considered. This may result in the solution space being disconnected or sparsely connected, possibly reducing solution quality. To overcome this problem, three means of reconnecting the solution space have been developed. The first widens the neighbourhood to include more diverse solutions. The second attempts to search disjoint components of the solution space individually, using a diversification function to guide the search. The final method reintroduces solutions that break the binding constraints to the solution space while penalising them in the cost function. Extending the neighbourhood using a graph theoretic entity called a Kempe chain significantly increases solution quality. Several explanations have been investigated. This research shows that a combination of a phased approach and Kempe chains forms the basis of an effective and flexible simulated annealing based examination timetabling system, which has been implemented successfully at a U.K. university.
Supervisor: Not available Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.639229  DOI: Not available
Share: