Comparison of static and dynamic penalty functions for handling constraints in genetic algorithm applied to course timetabling for CSM in UPMin /
Concepcion, Erick Castillo.
Comparison of static and dynamic penalty functions for handling constraints in genetic algorithm applied to course timetabling for CSM in UPMin / Erick Castillo Concepcion. - 2006 - 54 leaves
Thesis, Undergraduate (BS Applied Mathematics) -- U. P. in Mindanao
This study presents an application of generic algorithm (GA) on university course timetabling problem using penalty functions were utilized as fitness function, penalizing only on the soft constraints since all hard constraints must be satisfied. One is static penalty function having a fixed penalty parameter in the entire optimization process, and dynamic penalty function where the penalty factors are dependent on the current generation. In comparing the solution between the best for static penalty and the best for dynamic penalty, the later has relatively lower constraint violation making that solution better than the static penalty. Thus, dynamic penalty function is a better performer than static penalty function as a fitness function in a GA optimization process, although the soft constraints were partitioned into equality and inequality constraints, it is not clear which among the constraints is more violated because they have a different degree of penalty
Genetic algorithm.
Timetabling.
Timetbaling--Courses.
Static functions.
Dynamic functions.
Handling constraints.
Undergraduate Thesis --AMAT200
Comparison of static and dynamic penalty functions for handling constraints in genetic algorithm applied to course timetabling for CSM in UPMin / Erick Castillo Concepcion. - 2006 - 54 leaves
Thesis, Undergraduate (BS Applied Mathematics) -- U. P. in Mindanao
This study presents an application of generic algorithm (GA) on university course timetabling problem using penalty functions were utilized as fitness function, penalizing only on the soft constraints since all hard constraints must be satisfied. One is static penalty function having a fixed penalty parameter in the entire optimization process, and dynamic penalty function where the penalty factors are dependent on the current generation. In comparing the solution between the best for static penalty and the best for dynamic penalty, the later has relatively lower constraint violation making that solution better than the static penalty. Thus, dynamic penalty function is a better performer than static penalty function as a fitness function in a GA optimization process, although the soft constraints were partitioned into equality and inequality constraints, it is not clear which among the constraints is more violated because they have a different degree of penalty
Genetic algorithm.
Timetabling.
Timetbaling--Courses.
Static functions.
Dynamic functions.
Handling constraints.
Undergraduate Thesis --AMAT200