A Hybrid Ant Colony Optimization and Kernel Extreme Learning Machine Approach for Collaborative Transportation and Inventory Optimization in Cold Chain Logistics Based on ACO-KELM

Abstract

With the increasing demand for cold chain logistics for food, medicine, and other industries, how to improve the transportation efficiency and inventory management level of cold chain logistics has become a research hotspot. This paper proposes a collaborative optimization model of cold chain logistics transportation and inventory based on an ant colony optimization algorithm (ACO) and Kernel Extreme Learning Machine (KELM). The core of this model is to combine transportation route optimization with the forecasting function of inventory management, optimize the transportation route through ACO, and use KELM to accurately forecast inventory demand to realize the dual optimization of transportation and inventory. The comprehensive optimization of transportation routes, inventory holding cost, out-of-stock rate, and other objectives are considered by establishing the collaborative optimization objective function of the cold chain logistics system. This paper proposes a collaborative optimization model of cold chain logistics transportation and inventory based on an ant colony optimization algorithm (ACO) and Kernel Extreme Learning Machine (KELM). The model is validated on a real-world cold chain network with 5 distribution centers, 30 retailers, and dynamic demand data spanning 12 months. Experimental results show that the ACO-KELM model achieves a 18.7% reduction in total logistics cost (from 58.74 to 47.79) and a 22.3% lower out-of-stock rate (12.02% vs. 15.46%) compared to traditional single-objective optimization. The inventory demand forecasting via KELM demonstrates a mean absolute error (MAE) of 4.2 units, outperforming the support vector machine (SVM) baseline by 20.1% in prediction accuracy. The transportation energy consumption is reduced to 36.82, with the system's overall energy efficiency improved to 96.54%, indicating significant optimization in both cost and sustainability.