An Event-Based Parameterized Active Scheduler for Classical Job Shop Problem

摘要

In this work, an event-based genetic procedure for creating parameterized active schedules is proposed to solve the classical job shop scheduling, modelled as a continuous optimization problem. Instead of work with priorities values, the genetic algorithm defines values of delay times while the priorities are determined on the basis of the Last In First Out rule. The hypothesis is that any delay must end when the priority task arrives the machine. The scheduler is applied in a hybrid approach to solve the scheduling, which is a well-known NP-hard combinatorial optimization problem. After an initial schedule is created, it is refined by a local search and used as a seed in a final phase of optimization, in which a binary genetic indirect coding to induce permutations and generate new solutions is used. Preliminary results on a set of standard instances from literature validate the effectiveness of the proposed approach.