Enhancing Warehouse Efficiency With Time Series Clustering: A Hybrid Storage Location Assignment Strategy

Abstract

In the rapidly evolving domain of e-commerce, effective warehouse management emerges as a critical factor for ensuring timely deliveries. This paper addresses the Storage Location Assignment Problem (SLAP) in e-commerce warehouses, a challenge intensified by varying product volumes and unpredictable demands. We introduce a novel Intelligent Storage Location Assignment (ISLA) method that utilizes advanced time series clustering algorithms specifically, Self-Organizing maps, dynamic time warping-Based k-means, and Agglomerative Hierarchical Clustering (AHC), to optimize order fulfillment and enhance ware-house efficiency. By clustering and positioning items with similar demand patterns, our approach minimizes order preparation time, reduces unnecessary warehouse movements, and improves operational flows. Our empirical evaluation, based on a real-world dataset from Kaggle, demonstrates the superiority of AHC in efficiently grouping high-turnover items, as evidenced by higher silhouette scores. Applying this method in simulations across various picking strategies such as s-shape, mid-point, discrete order picking, zone picking, and batch picking, we achieve significant efficiency improvements. Notably, our ISLA method results in up to 61% and 69% efficiency gains under s-shape and midpoint routing policies, respectively, outperforming traditional random and ABC storage assignments. These results not only highlight the significant potential of Artificial Intelligence (AI) in revolutionizing warehouse operations but also bridge the existing knowledge gap by showcasing a practical and impactful application of AI in SLAP. Our research advances the field of smart logistics, emphasizing the critical role of AI-driven intelligent storage location assignment in optimizing warehouse processes and enhancing the efficiency of the e-commerce supply chain.