Optimizing E-commerce Logistics with ARIMA-BP Neural Networks and GA: A Multi-Objective Approach for Emergency Response and Network Efficiency

In the digital era, the rapid development of e-commerce logistics networks is crucial for maintaining operational efficiency, especially under emergency conditions such as epidemics and earthquakes. This study introduces a novel framework for optimizing e-commerce logistics by integrating AutoRegressive Integrated Moving Average (ARIMA) with Back Propagation (BP) neural networks, further refined through Genetic Algorithm (GA) optimization. Utilizing cargo volume data over a two-year period, we developed a sophisticated ARIMA-BP model to forecast the daily cargo volume for each route for the subsequent month. The subsequent application of a GA-optimized BP neural network aims to achieve a multi-objective optimization, focusing on minimizing the impact of route disruptions and balancing workload distribution across the network. We employed network centrality measures—degree, closeness, and betweenness—to evaluate the significance of logistics sites and routes, enhancing the network's resilience and operational efficiency. This approach also facilitated the integration of new logistics sites and routes, bolstering network performance. Further, we introduced a stochastic perturbation technique to explore the network's adaptability to both anticipated and unforeseen flow changes post-route closure. Finally, complex network topology analysis was applied to assess the robustness of the logistics network. Our methodology exemplifies a comprehensive strategy for e-commerce logistics optimization, offering a robust framework for enhancing network resilience and efficiency in face of emergencies, thereby contributing to the reduction of operational costs and improvement of service quality.