An Evolutionary Variable Neighbourhood Search for the Unrelated Parallel Machine Scheduling Problem

This article addresses a challenging industrial problem known as the unrelated parallel machine scheduling (UPMSP) with sequence-dependent setup times. In UPMSP, we have set of machines and group jobs. The goal is to find optimal way schedule jobs for execution by one several available machines. UPMSP has been classified an NP-hard optimisation and, thus, cannot be solved exact methods. Meta-heuristic algorithms are commonly used sub-optimal solutions. However, large-scale instances pose significant challenge meta-heuristic algorithms. To effectively solve this introduces two-stage evolutionary variable neighbourhood search (EVNS) methodology. proposed EVNS integrates algorithm descent framework in adaptive manner. employed first stage. It uses mix crossover mutation operators generate diverse second stage, propose exploit area around solutions generated A dynamic strategy developed determine switching time between these two stages. guide towards promising areas, diversity-based fitness function explore different locations landscape. We demonstrate competitiveness presenting computational results comparisons on 1640 benchmark instances, which literature. experiment show that our obtains better than compared instances.