Trolleyler ile yük taşima ve AGV rotalama için uyarlamali büyük komşuluk arama algoritmasi

Automated Guided Vehicles (AGVs) play a critical role in industrial logistics, providing efficient material handling solutions. This study addresses a real-world problem of AGV routing and load handling at Vestel Electronics, specifically focusing on their refrigerator factory, which produces a diverse range of models, including no-frost refrigerators, minibars, and built-in refrigerators. The diversity in refrigerator models necessitates the transportation of various materials such as refrigerator doors, compressors, and shelves within strict delivery time windows. Given the limited number of AGVs and trolleys available to increase their load capacity, determining the optimal routing and scheduling of pickups and deliveries presents a complex challenge. To address this, we develop a Mixed Integer Linear Programming (MILP) model to formalize the AGV routing and load handling problem, incorporating trolley allocation and aiming to minimize total costs, including travel, earliness, and tardiness penalties. The model also considers practical constraints like time windows and load capacities, making it suitable for real-world applications. However, due to the computational inefficiency of the MILP model in solving medium to large-sized instances, we propose an Adaptive Large Neighborhood Search (ALNS) algorithm. This algorithm, with its adaptive mechanisms, efficiently explores diverse neighborhood structures, finding near-optimal solutions. Extensive computational experiments show that the ALNS algorithm is robust across varying instance sizes, delivering efficient solutions and reducing computation times significantly. A real-life case study at Vestel Electronics highlights the approach's potential, reducing operational costs to as low as 1/48th of the current costs, thus offering substantial cost savings and operational efficiency.