Abstract
An NP-hard problem like Flexible Job Shop Scheduling (FJSP) tends to be more complex and requires more computational effort to optimize the objectives with contradictory measures. This paper aims to address the FJSP problem with combined and contradictory objectives, like minimization of make-span, maximum workload, and total workload. This paper proposes ‘Hybrid Adaptive Firefly Algorithm’ (HAdFA), a new enhanced version of the classic Firefly Algorithm (FA) embedded with adaptive parameters to optimize the multi objectives concurrently. The proposed algorithm has adopted two adaptive strategies, i.e., an adaptive randomization parameter (α) and an effective heterogeneous update rule for fireflies. The adaptations proposed by this paper can help the optimization process to strike a balance between diversification and intensification. Further, an enhanced local search algorithm, Simulated Annealing (SA), is hybridized with Adaptive FA to explore the local solution space more efficiently. This paper has also attempted to solve FJSP by a rarely used integrated approach where assignment and sequencing are done simultaneously. Empirical simulations on benchmark instances demonstrate the efficacy of our proposed algorithms, thus providing a competitive edge over other nature-inspired algorithms to solve FJSP.
Highlights
Flexible Job Shop Scheduling (FJSP) differs from the classic job shop scheduling problem in being flexible
The results show that the Hybrid Adaptive Firefly Algorithm’ (HAdFA) performs better than Adaptive Firefly Algorithm (FA) for 8 instances with p-value < 0.0001 except MK01, MK03 and MK08 for which p-value is ‘0‘ that indicates there is no significant difference in both algorithms
If we make the direct comparison between BEG-NSGA IId and HAdFAg, coded in Matlab on an Intel processor, it can be perceived that HAdFA takes just 0.3–2.252 s to compute for all MK problems
Summary
Flexible Job Shop Scheduling (FJSP) differs from the classic job shop scheduling problem in being flexible. Suitable machines have to be assigned to each operation, and those operations have to be sequenced to process These sub-problems can be addressed by two methods, i.e., a hierarchical and an integrated approach. Li et al [7] developed a two-tuple scheme to solve FJSP by an integrated approach They fine-tuned the control parameters for their proposed adaptive evolutionary algorithm. Kato et al [8] solved multi-objective FJSP They used Particle Swarm Optimization (PSO) for the assignment subproblem and Random Restart Hill Climbing (RRHC) for the scheduling subproblem, and they found the obtained results to be conclusive. Rohaninejad et al [22] solved the capacitated Job Shop Problem by Tabu search and FA following hierarchical and integrated approaches to minimize tardiness and overtime cost.
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
