Self-fertilization based genetic algorithm for university timetabling problem

In this paper, a new algorithm inspired from the self-fertilization of some plants is proposed for the university timetabling problem (UTP). The main idea of the algorithm is to modify the fitness function, the selection and crossover operators of GA to obtain a further fit for UTP. Fitness function of this algorithm will neglect hard constraints because no infeasible individual can pass the check of advisor to survive. The advisor based on heuristic methods can also simplify the computation once there are changes on constraints. Distinguished from traditional crossover, a new exchange in one chromosome rather than between chromosomes will be issued to keep the integrity of the schedule. During some processes, simulated annealing was introduced as a select strategy for diversity of the population. This algorithm was implemented and tested with the real data of Tianjin University, China. The algorithm produces good timetable for the students and teachers and improve the usage rate of classroom. The experiment results indicate that our new hybrid genetic algorithm that addressing timetabling problem is promising and converge rapidly.