Smart Logistics Model for Supply Chain Management via Brain-Inspired Geometric Deep Networks

Area of Science:

• Supply Chain Management
• Artificial Intelligence
• Biomimetic Computing

Background:

• Intelligent logistics models are crucial for agile, sustainable, and resilient supply chains.
• Brain-inspired deep learning architectures like LSTM, GNN, and CNN offer advanced decision-making capabilities.
• These models are biomimetically inspired by biological information processing.

Purpose of the Study:

• To propose a novel hybrid deep learning strategy for smart supply chain logistics management.
• To leverage biomimetic computational principles for enhanced logistics decision-making.
• To improve supply chain agility, sustainability, and resilience through intelligent management.

Main Methods:

• A hybrid deep learning strategy combining LSTM, convolutional layers, and GraphSAGE geometric layers.
• Utilizing biomimetic particle swarm optimizer and Adam (PSO-Adam) for sequential optimization.
• Leveraging GraphSAGE for scalable graph learning and enhanced predictive accuracy on unseen data.

Main Results:

• Achieved high average accuracies (96.6%–100%) across five diverse logistics datasets.
• Demonstrated effectiveness in multi-category logistics parameter forecasting.
• Confirmed the model's potential for complex supply chain decision-making.

Conclusions:

• The proposed hybrid deep learning model offers a cost-efficient solution for intelligent logistics.
• The model enhances supply chain visibility, customer satisfaction, and industry reputation.
• Biomimetic geometric networks show significant potential for optimizing complex supply chain operations.