Performance evaluation of particles coding in particle swarm optimization with self-adaptive parameters for flexible job shop scheduling problem

Abstract

The metaheuristic Particle Swarm Optimization (PSO) is well suited to solve the Flexible Job Shop Scheduling Problem (FJSP), and a suitable particle representation should importantly impact the optimization results and performance of this algorithm. The chosen representation has a direct impact on the dimension and content of the solution space. In this paper, we intend to evaluate and compare the performance of two different variants of PSO with different particle representations (PSO with Job-Machine coding Scheme (PSO-JMS) and PSO with Only-Machine coding Scheme (PSO-OMS)) for solving FJSP. These procedures have been tested on thirteen benchmark problems, where the objective function is to minimize the makespan and total workload and to compare the run time of the different PSO variants. Based on the experimental results, it is clear that PSO-OMS gives the best performance in solving all benchmark problems.