Scholarlink

<p>The rapid expansion of e-commerce has increased pressure on logistics systems to generate fast, cost-efficient, and adaptive delivery routes. Dynamically, traditional routing methods struggle to respond to fluctuating order volumes, traffic density, and resource limitations. To address these challenges, this research presents a Genetic Algorithm-optimized Attention-Based Bi-Directional Long Short-Term Memory (GA-ABi-LSTM) framework for intelligent route planning in urban delivery environments. The dataset comprises 1000 real and simulated delivery records collected from multiple urban distribution centers, which contain key operational features such as order coordinates, delivery windows, vehicle capacity, traffic indices, fuel costs, and route performance metrics. Pre-processing uses normalization and anomaly detection to ensure consistency, while PCA highlights the most influential routing determinants. A Genetic Algorithm (GA) evolves route candidates by optimizing both cost and time through selection, crossover, and mutation. Bi-LSTM captures forward and backward temporal dependencies in delivery sequences, enabling accurate modeling of time-varying logistics behavior. Attention Mechanism (AM) prioritizes high-impact delivery nodes, traffic-sensitive segments, and urgent orders to enhance decision quality. The model is implemented using Python, integrating GA for search optimization, Bi-LSTM for sequential learning, and AM for feature prioritization. Experimental results validate the superiority of the proposed system. GA-ABi-LSTM achieves an F1-Score of 0.912480, training accuracy of 0.953720, reduced computation time to 48.317&#xa0;s, and improved route metrics, including path length (185&#xa0;km) and completion time (125&#xa0;min) with a punctuality rate of 92%. Overall, the GA-ABi-LSTM framework significantly improves delivery efficiency, adaptability, and operational reliability compared with conventional logistics routing approaches.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Deep learning and genetic algorithm-driven e-commerce logistics route planning method

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月15日

查看全文
  • Jiping Liu

摘要

The rapid expansion of e-commerce has increased pressure on logistics systems to generate fast, cost-efficient, and adaptive delivery routes. Dynamically, traditional routing methods struggle to respond to fluctuating order volumes, traffic density, and resource limitations. To address these challenges, this research presents a Genetic Algorithm-optimized Attention-Based Bi-Directional Long Short-Term Memory (GA-ABi-LSTM) framework for intelligent route planning in urban delivery environments. The dataset comprises 1000 real and simulated delivery records collected from multiple urban distribution centers, which contain key operational features such as order coordinates, delivery windows, vehicle capacity, traffic indices, fuel costs, and route performance metrics. Pre-processing uses normalization and anomaly detection to ensure consistency, while PCA highlights the most influential routing determinants. A Genetic Algorithm (GA) evolves route candidates by optimizing both cost and time through selection, crossover, and mutation. Bi-LSTM captures forward and backward temporal dependencies in delivery sequences, enabling accurate modeling of time-varying logistics behavior. Attention Mechanism (AM) prioritizes high-impact delivery nodes, traffic-sensitive segments, and urgent orders to enhance decision quality. The model is implemented using Python, integrating GA for search optimization, Bi-LSTM for sequential learning, and AM for feature prioritization. Experimental results validate the superiority of the proposed system. GA-ABi-LSTM achieves an F1-Score of 0.912480, training accuracy of 0.953720, reduced computation time to 48.317 s, and improved route metrics, including path length (185 km) and completion time (125 min) with a punctuality rate of 92%. Overall, the GA-ABi-LSTM framework significantly improves delivery efficiency, adaptability, and operational reliability compared with conventional logistics routing approaches.