Parallelism and Hybridization in Differential Evolution to solve the Flexible Job Shop Scheduling Problem

Abstract

The Flexible Job Shop Scheduling Problem (FJSSP) is one of the most challenging combinatorial optimization problems, with practical applicability in a real production environment. In this work, we propose a simple Differential Evolution (DE) algorithm to tackle this problem. To represent a FJSSP solution, a real value representation is adopted, which requires a very simple conversion mechanism to obtain a feasible schedule. Consequently, the DE algorithm still works on the continuous domain to explore the problem search space of the discrete FJSSP. Moreover, to enhance the local search ability and to balance the exploration and exploitation capabilities, a simple local search algorithm is embedded in the DE framework. Also, the parallelism of the DE operations is included to improve the efficiency of the whole algorithm. Experimental results confirm the significant improvement achieved by integrating the modifications introduced in this study. Additionally, test results show that our algorithm is competitive when compared with most existing approaches for FJSSP.