Hybrid Algorithm to Solve Timetabling Problem

Abstract

The university course timetabling problem is a well-known highly-constrained difficult optimization problem. The problem seeks the best allocation of courses to time slots and rooms while ensuring all related constraints are satisfied. Due to the limited resources (rooms and time slots), finding an optimal, or even a high quality, timetable is a challenging task that every university encounters every semester. Many metaheuristic algorithms have been proposed for university course timetabling problem. Genetic algorithm is a class of metaheuristic and has shown very good results for many real-world problems. However, for university course timetabling problems, a traditional genetic algorithm is not usually considered as an efficient solver because it is very hard to maintain the solution feasibility. In this research, we propose a new hybrid algorithm that combines genetic algorithm with simulated annealing to find good solutions for university course timetabling problems. The proposed hybrid algorithm uses simulated annealing in adaptive manner to rectify solutions and to improve the quality of the generated solution by genetic algorithm. The proposed algorithm is tested over Socha dataset from the scientific literature and compared with the state of the art methods.