Scholarlink

<p>As the logistics sector undergoes rapid digital transformation, efficient and adaptive delivery route planning is essential to guarantee prompt delivery and process cost reduction. This research proposes a framework that utilizes Deep Learning (DL) and Neural Network Optimization of Logistics Delivery Routes to optimize routing in dynamic logistics. The proposed framework is driven by multi-source operational data utilized in the Logistic Operations &amp; Risk Dataset and is executed through a preprocessing pipeline of data management, which includes missing data management and scaled normalization to develop a standardized dataset. Additionally, a feature extraction technique, specifically Principal Component Analysis (PCA), was employed to reveal the primary modes of mobility and operations. The approach proposed for logistics delivery routes is the Namib Beetle Optimized Spatio-Temporal Recurrent Neural Network (NBO-ST-RNN). Specifically, the NBO system employs a stochastic modeling process to learn optimal configurations of routes utilizing real-time data-sourced information autonomously. The performance evaluation demonstrates that this approach results in a decrease in Mean Absolute Error (MAE) to 9845.23, a reduction in Mean Absolute Percentage Error (MAPE) to 0.073, and an increase in R-squared (R<sup>2</sup>) to 0.9875, confirming prediction accuracy. Furthermore, the proposed model exhibited a 91.3% vehicle utilization rate, a reduction in operating cost of a magnitude of 13.5%, and an increase in customer satisfaction at 98.6%, confirming practical implications. Enabling data-driven, real-time decision-making, this research offers a scalable and intelligent strategy for optimization that contributes to improved resilience, sustainability, and performance in logistics delivery networks through smart transportation systems.</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 neural network-based optimization of logistics delivery routes

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2026年5月28日

查看全文
  • Xijing Ou

摘要

As the logistics sector undergoes rapid digital transformation, efficient and adaptive delivery route planning is essential to guarantee prompt delivery and process cost reduction. This research proposes a framework that utilizes Deep Learning (DL) and Neural Network Optimization of Logistics Delivery Routes to optimize routing in dynamic logistics. The proposed framework is driven by multi-source operational data utilized in the Logistic Operations & Risk Dataset and is executed through a preprocessing pipeline of data management, which includes missing data management and scaled normalization to develop a standardized dataset. Additionally, a feature extraction technique, specifically Principal Component Analysis (PCA), was employed to reveal the primary modes of mobility and operations. The approach proposed for logistics delivery routes is the Namib Beetle Optimized Spatio-Temporal Recurrent Neural Network (NBO-ST-RNN). Specifically, the NBO system employs a stochastic modeling process to learn optimal configurations of routes utilizing real-time data-sourced information autonomously. The performance evaluation demonstrates that this approach results in a decrease in Mean Absolute Error (MAE) to 9845.23, a reduction in Mean Absolute Percentage Error (MAPE) to 0.073, and an increase in R-squared (R2) to 0.9875, confirming prediction accuracy. Furthermore, the proposed model exhibited a 91.3% vehicle utilization rate, a reduction in operating cost of a magnitude of 13.5%, and an increase in customer satisfaction at 98.6%, confirming practical implications. Enabling data-driven, real-time decision-making, this research offers a scalable and intelligent strategy for optimization that contributes to improved resilience, sustainability, and performance in logistics delivery networks through smart transportation systems.