Penjadwalan Mesin Produksi Biji Kopi Menggunakan Hybrid Particle Swarm Optimization Dan Variabel Neighborhood Search

Abstract

The coffee processing industry, especially at the home industry scale, faces challenges in optimizing production machine scheduling to minimize total completion time (makespan) and improve efficiency. This research aims to analyze the performance and generate an optimal scheduling model for coffee bean production machines in a flow shop environment by implementing a hybrid algorithm of Particle Swarm Optimization (PSO) and Variable Neighborhood Search (VNS). The hybrid method is designed to leverage PSO's strength in global exploration and VNS's capability in local exploitation. Research data were obtained from observations of the production process at RKB Roastery, including processing times for the preheating, roasting, cooling, grinding, and sealing stages. Algorithm implementation involved extensive parameter testing to determine the best configuration. Experimental results show that the hybrid PSO-VNS algorithm with parameters inertia weight (w=0.6), cognitive coefficient (c1=2.0), social coefficient (c2=2.0), and neighborhood (k=5) successfully produced the lowest makespan of 91704.69 time units. This value is significantly superior compared to the makespan from standard PSO (110308.96) and standard VNS (477628.23). Therefore, this study concludes that the hybrid PSO-VNS approach is effective in optimizing coffee bean production machine scheduling, offering a better solution for minimizing total production time and potentially improving resource utilization.