Analyzing the Effect of Random-Key Optimizer Agents on Parallel Machine Sequencing Problem

This study analyzes the isolated and cooperative effect of Random-Key Optimizer (RKO) agents on the unrelated parallel machine scheduling problem with sequence-dependent setup times (UPMSP-SDS) to minimize makespan. A heuristic-free decoder was used to isolate the native search capacity. Ablation experiments on instances with 200 jobs (10 to 30 machines) indicate that the triparadigm cooperative orchestration (GA+VNS+ILS) yields synergistic gains, achieving the lowest mean gap and mitigating the early stagnation of solutions.