Hybrid Genetic Algorithms for Part Type Selection and Machine Loading Problems with Alternative Production Plans in Flexible Manufacturing System

Abstract

This paper addresses two NP-hard and strongly related problems in production planning of flexible manufacturing system (FMS), part type selection problem and machine loading problem. Various flexibilities such as alternative machines, tools, and production plans are considered. Real coded genetic algorithms (RCGA) that uses an array of real numbers as chromosome representation is developed to handle these flexibilities. Hybridizing with variable neighbourhood search (VNS) is performed to improve the power of the RCGA exploring and exploiting the large search space of the problems. The effectiveness of this hybrid genetic algorithm (HGA) is tested using several test bed problems. The HGA improves the FMS effectiveness by considering two objectives, maximizing system throughput and minimizing system unbalance. The resulted objective values are compared to the optimum values produced by branch-and-bound method. The experiments show that the proposed RCGA could reach near optimum solution and the hybridization can improve the performance of the RCGA.