Multi-Objective Evolutionary Algorithm for University Class Timetabling Problem

After their successful application to a wider range of prob- lems, in recent years evolutionary algorithms (EAs) have also been found applicable to many challenging problems, like complex and highly con- strained scheduling problems. The inadequacy of classical methods to handle discrete search space, huge number of integer and/or real vari- ables and constraints, and multiple objectives, involved in scheduling, have drawn the attention of EAs to those problems. Academic class timetabling problem, one of such scheduling problems, is being studied for last four decades, and a general solution technique for it is yet to be formulated. Despite multiple criteria to be met simultaneously, the prob- lem is generally tackled as single-objective optimization problem .M ore- over, most of the earlier works were concentrated on school timetabling, and only af ew on university class ti metabling. On the other hand, in many cases, the problem was over-simplified by skipping many complex class-structures. The authors have studied the problem, considering dif- ferent types of class-structures and constraints that are common to most of the variants of the problem. NSGA-II-UCTO, a version of NSGA-II (an EA-based multi-objective optimizer) with specially designed repre- sentation and EA operators, has been developed to handle the problem. Though emphasis has been put on university class timetabling, it can also be applied to school timetabling with a little modification. The suc- cess of NSGA-II-UCTO has been presented through its application to two real problem sf ro m a technical institute in India.