Learning Scheduling Rule for Job Shop Considering Multi-objective Evaluation Related to Due Date

Abstract

This paper deals with job shop scheduling problem with consideration for multi-objective evaluation related to due dates. An effective priority rule is constructed for this problem using a neural network. The neural network is trained using input-output pairs extracted from schedules optimized using a genetic algorithm incorporating the priority rule. The numerical experiments demonstrate that effective priority rules can be automatically constructed by training using the input-output pairs.