摘要
随着全球经济一体化和“一带一路”倡议的推进,港口物流在区域经济发展中的作用愈发重要。然而,客户对物流服务需求的多样化和个性化,使得传统物流方案决策系统难以满足复杂需求,制约了港口物流企业的竞争力。研究提出了一种基于客户需求的多式联运物流方案推荐算法(COMLRA),通过深度学习技术构建智能推荐决策系统,精准识别客户需求并生成个性化物流方案。该算法利用嵌入层、注意力机制、交叉网络和深度网络,捕捉特征间的复杂交互关系,并引入动态损失权重和排序损失优化预测性能。实验结果表明,COMLRA在HR@5和NDCG@5指标上优于其他方法,验证了其有效性和优越性。该研究为港口物流企业提供了高效、灵活的决策支持工具,助力智慧物流技术的深度应用和港口物流行业的智能化升级。
With the promotion of global economic integration and the“the Belt and Road”initiative,port logistics plays an increasingly important role in regional economic development.However,the diversification and personalization of customer needs for logistics services make it difficult for traditional logistics decision-making systems to meet complex needs,which constrains the competitiveness of port logistics enterprises.This study proposes a customer need based multimodal logistics solution recommendation algorithm(COMLRA),which constructs an intelligent recommendation decision system through deep learning technology to accurately identify customer needs and generate personalized logistics solutions.This algorithm utilizes embedding layers,attention mechanisms,crossover networks,and deep networks to capture complex interactions between features,and introduces dynamic loss weights and ranking losses to optimize prediction performance.The experimental results indicate that COMLRA is HR@5 and NDCG@5.It outperforms other methods in terms of indicators,verifying its effectiveness and superiority.This study provides efficient and flexible decision support tools for port logistics enterprises,facilitating the deep application of smart logistics technology and the intelligent upgrading of the port logistics industry.
作者
仲兆满
徐俊康
李梦晗
陈柯含
ZHONG Zhaoman;XU Junkang;LI Menghan;CHEN Kehan(School of Computer Engineering,Jiangsu Ocean University,Lianyungang 222005,China;Jiangsu Research Institute of Marine Resources Development,Lianyungang 222005,China)
出处
《江苏海洋大学学报(自然科学版)》
2026年第1期54-65,共12页
Journal of Jiangsu Ocean University:Natural Science Edition
基金
国家自然科学基金资助项目(72174079)
江苏省“青蓝工程”大数据优秀教学团队(2022-29)
连云港市重点研发(产业前瞻与关键核心技术)项目(CG2323)。
关键词
客户需求
推荐算法
多式联运
注意力机制
深度学习
customer needs
recommendation algorithm
multimodal transport
attention mechanism
deep learning
