Combined Multi-Agent Method to Control Inter-Department Common Events Collision for University Courses Timetabling

Abstract

University course timetabling is the scheduling of courses at different time slots in a university. The two important issues in this process are (i) the allocation of all events (professors, courses, and students) to resources (time slots daily/weekly and theory/practical classes) in a semester, and (ii) maximizing the satisfaction of common events (professors, courses, and students) among multiple departments. Accumulating evidences in university course timetabling problems suggest dividing the problem into several sub-problems. This study attempted to investigate the appropriateness of using the genetic algorithm (GA) and the imperialist competitive algorithm (ICA). The proposed technique consists of two steps: (i) using the proposed manipulated GA for solving the timetabling problem of each department, and (ii) eliminating the interference of common events among multiple departments and satisfying the hard and soft constraints by using ICA. Finally, a report on the efficiency of the methodology used in this study was obtained from the University of Tabriz in Iran and University of Udine in Italy. In this paper, the results are revealed in two ways: (i) reduction in the problems due to shrinking of the database and solving of the problems in parallel and (ii) solving the different parts of the problem by using various criterion results, increasing the common events satisfaction in that sub-problem. Eventually, the proposed model provided successful satisfaction of the hard constraints in <700 iterations with GA and elimination of interference in 40 iterations with ICA in most of the cases.

Keywords

• university course timetabling
• imperialist competitive algorithm
• genetic algorithm