A Hybrid Unsupervised Learning Framework for Centralized Kitchen Food Supply Optimization

With the rapid development of information and intelligent technologies, food supply systems for centralized kitchens face several challenges. These include large-scale order volumes, strict delivery time constraints, and high scheduling complexity. Traditional methods that rely on manual experience are insufficient to ensure efficient and reliable delivery operations. To address these challenges, this paper designs and implements a food supply system for centralized kitchens (FSSCK). This system supports integrated order management, inventory coordination, and delivery scheduling. At the delivery decision level, this paper proposes a delivery scheduling method based on unsupervised clustering and time-aware scheduling. First, it applies k-means clustering to capture the spatial distribution of orders. Then, it constructs a scheduling strategy that jointly considers delivery deadlines and geographic proximity to determine the delivery sequence. The proposed system improves scheduling efficiency and operational coordination in time-sensitive distribution scenarios. The results demonstrate its practical value for enhancing the informatization level and management efficiency of centralized kitchen food supply systems.