A hybrid genetic algorithm for solving the joint batching and scheduling of a batch processor with earliness and tardiness penalties

Abstract

In this article, we address the problem of minimising the total weighted earliness and tardiness penalties of a batch processor by integrating genetic algorithms and math programming for determining the allocation of the customer orders to production batches and scheduling the resulting batches. Each job has its own unique due date, and earliness and tardiness penalties. A genome representation is introduced for solving the scheduling problem and is evolved by a genetic algorithm while at each evolution, the genome score is evaluated by a mathematical program for determining the job size per batch and the formation of batches. The genetic algorithm's performance is compared with solutions found by a non-linear integer math program solver and its linearised model proposed by Dessouky, Kijowski and Verma (1999) on a set of representative test problems. The developed hybrid genetic algorithm proves its capability and superiority to find good solutions for the problem under consideration and outperforms solutions from a commercial optimisation package, CPLEX.