Genetic Algorithm and Their Applications to Scheduling Hyungjun Park
Introduction Based on the ideas of evolution and natural selection Pioneered by John Holland in 1960’s
Biological Background Genes Natural Selection and Fitness Reproduction Crossover Mutation
Genetic Algorithm Start Repeat Until End Condition End
Traveling Salesman Problem [Start] Random routes + heuristic [Fitness] Inverse of distance [Selection] Roulette wheel selection [Crossover] Leave blanks then refill [Mutation] Interchange Parameter Value Population Size 1,000 Mutation 3% Elitism Yes End Condition 200 Iterations
GA Performance Weighted Tardiness Descent Methods AU DES, DESO GA outperforms # of jobs DES DESO GA 50 20.78% 21.02% 33.68% 100 19.70% 19.47% 33.71% 200 17.46% 17.70% 33.72% 500 14.89% 14.81% 31.76%
Q & A