Dynamic Vehicle-Cargo Matching Based on Adaptive Time Windows

The core task of vehicle-cargo matching is to dispatch the cargoes to the trucks. The existing matching policies mainly focus on maximizing the shipping weight for each truck. Due to each cargo is bulky and heavy in bulk logistics area, such strategies cannot ensure maximization of total weight of cargoes to be transported, and lead to a few cargoes be stranded. To tackle this issue, we present an intelligent decision framework for vehicle-cargo matching, called ILPD. Based on the limiting rules and features related to loading plan decisions that extracted from historical logistics data, we design a time window-based matching policy to achieve the goal of maximizing the total shipping weight and minimizing the quantity of stranded cargoes. Specifically, in each time window, dynamic programming and Branch-and-Bound method are leveraged to generate the loading plans of cargoes with the aim of minimization of stranded cargoes’ quantities. Then, Kuhn-Munkres algorithm is used to make the matching decisions to obtain maximum weight matching. Further, to fit for dynamic changing number of trucks and cargoes, a time zone-based Q-learning algorithm is proposed to adjust the time window size adaptively. Extensive experimental results on real data sets validate the effectiveness and practicality of our proposal.