Hybrid frameworks for flexible job shop scheduling

Effective scheduling is essential for manufacturing firms to survive in today's fiercely competitive marketplace. Improvisation of schedule by simultaneous optimization of the performance measures is imperative for the manufacturing firms to stay ahead of competition and is one of the key responsibilities of shop floor managers. The current study addresses flexible job shop scheduling problem (FJSSP), considering three objectives: makespan (MS), maximal machine workload (MW), and total workload (TW). This research approaches the problem by developing two hybrid frameworks. The first framework (Hybrid-I) is a co-evolutionary combination of the social thinking capability of particle swarm optimization and local search capability of gravitational search algorithm. The second approach (Hybrid-II), hybridize the effectiveness of genetic algorithm in finding global best region with PSO's cluster interactions to improve the search for an optimal solution. A well-designed, efficient version of PSO (ePSO), that inherits twofold improvement through variable random function strategy and mutation strategy, is applied in both the approaches. With the view to make a reasonable comparison between both approaches and with the state-of-the-art methods, tests have been conducted on 28 benchmark instances taken from three different data sets. Further, the managerial implication of this research has been validated by implementing Hybrid-II for an industrial case. The substantial performance of the proposed approach on benchmark instances as well as real-life industrial data supports a strong candidature for optimization of FJSSP.