山东交通学院学报

为提高网络货运平台车货匹配效率，考虑车方、货方和网络货运平台三方需求设计三阶段车货供需匹配模型，第一阶段建立货物属性分类模型，识别并消除不同属性货物间的冲突关系；第二阶段建立车辆信息筛选模型，提取车辆的高频配送区域；第三阶段建立多目标匹配模型，以车方利润最大化、货方成本最小化和网络货运平台收益最大化为目标建立目标函数。引入贪心算子、Circle混沌映射、反向学习策略及模拟退火机制改进传统遗传算法，用于求解模型。以中国物通网公布的广州出发货源及车源信息为例对三阶段车货供需匹配模型及其求解算法开展实证分析。结果表明：采用改进遗传算法求解得到的车货匹配方案在整体装载效率与车辆利用率方面均优于传统遗传算法，在相同的货物总量下，改进遗传算法使用更少的车辆完成运输任务，平均每辆车装载的货物更多，装载集中度显著提高；改进遗传算法收敛速度快，收敛后曲线更平稳，波动较小；相较于传统遗传算法，改进遗传算法匹配方案的车方利润在4个货物集合分别增加约32.6%、40.6%、48.6%、31.3%,货方成本相应减少约3.2%、10.3%、11.5%、7.6%,平台收益分别增加约0.9%、0.1%、2.4%、0.1%,改进遗传算法在降低货方成本的基础上，显著提高车方利润，并实现网络货运平台收益微增。构建的三阶段车货供需匹配模型及改进遗传算法在统筹车方、货方与网络货运平台三方利益方面表现更优。

To improve vehicle-cargo matching efficiency on online freight platforms, a three-stage supply-demand matching model is proposed that accounts for the needs of carriers, shippers, and the platform. Stage 1 constructs a cargo attribute classification model to identify and eliminate conflicts among different cargo attributes. Stage 2 develops a vehicle information screening model to extract each vehicle′s high-frequency delivery regions. Stage 3 formulates a multi-objective matching model with objective functions that maximize carrier profit, minimize shipper cost, and maximize platform revenue. A traditional genetic algorithm is enhanced with a greedy operator, a circle chaotic map, an opposition-based learning strategy, and a simulated annealing mechanism, and the enhanced algorithm is used to solve the model. An empirical analysis is conducted using Guangzhou-origin freight and vehicle postings released by the China Wutong Network. The results indicate that the matching solutions obtained with the improved genetic algorithm outperform those from the standard genetic algorithm in overall loading efficiency and vehicle utilization. Given the same total shipment volume, the improved algorithm completes transport tasks with fewer vehicles, increases the average number of loads per vehicle, and significantly improves load consolidation. It also converges faster and exhibits a smoother, less volatile post-convergence curve. Compared with the standard genetic algorithm, the improved algorithm increases carrier profit by approximately 32.6%, 40.6%, 48.6%, and 31.3% across four cargo sets, reduces shipper cost by about 3.2%, 10.3%, 11.5%, and 7.6%, and raises platform revenue by about 0.9%, 0.1%, 2.4%, and 0.1%, respectively. Overall, the three-stage matching model and the improved genetic algorithm better balance the interests of carriers, shippers, and online freight platforms.