2104 ニューラルネットワークを用いた納期遅れコストと段取り時間の二目的スケジューリング : SA による学習

概要

論文の詳細を見る
This report proposes a scheduling method for a bi-criteria job shop scheduling problem. The criteria considered in this report are tardiness cost and sequence-dependent setup time. Two priority rules each of which is effective to each scheduling criterion are used to make a composite dispatching rule. One of the two rules is SIMSET rule that selects the next job with the shortest setup time SIMSET rule can make effective schedules with regard to average setup time. The other rule is constructed using a neural network. This neural network is trained by a simulated annealing method to make effective schedules with regard to tardiness cost. The proposed composite rule has the parameter α and can make effective bi-criteria scheduling by tuning it. Numerical experiments show that the effectiveness of the proposed rule.
一般社団法人日本機械学会の論文
2003-03-28