关键词: 交通运输工程；需求预测；前摄性配送；混合模拟遗传算法；两级路网；电动汽车

Abstract: Under the requirement of massive, dispersed, and randomly arrived e-commerce logistics orders, a proactive logistics service mode based on demand prediction was proposed. The product attributes, user behaviors, customer interaction characteristics and other behavior trajectory data were combined to predict the potential purchasing intentions of customers, and the predicted order demands was advanced to customers’ nearest warehouse. Therefore, a two-echelon electric vehicle network optimization model of “central depot to forward center to customer” was established. Given the complexity of the problem, a two-stage hybrid heuristic intelligent optimization algorithm was designed. In the first stage, a two-edge successive correction method was used to obtain the initial solution, and in the second stage, a hybrid simulated genetic algorithm was designed to iteratively optimize the initial solution. The experimental results show that the total transportation distance and total cost of the first level network for proactive delivery are reduced by 106% and 101%, respectively, while the total cost of the second level network is increased by only 9%.

Key words: traffic and transportation engineering; demand forecasting; proactive delivery; hybrid simulation genetic algorithm; two-echelon road network; electric vehicle