Supply chain network based on Block Chain and deep learning

Abstract

Recent studies have explored blockchain technology and Graph Neural Networks (GNNs) in supply chain management; however, most approaches focus on isolated tasks such as demand forecasting or supplier evaluation without integrating intelligent decision-making and secure transaction verification within a unified framework. This paper proposes an intelligent supply chain management framework that combines Graph Neural Networks (GNNs), Multi-Agent Systems (MAS), and blockchain technology to improve prediction accuracy, operational transparency, and decision reliability. The proposed model represents suppliers, warehouses, inventory units, and logistics routes as interconnected graph structures to capture complex relational dependencies. Intelligent agents support inventory forecasting, supplier evaluation, and logistics planning, while blockchain technology ensures secure and tamper-resistant transaction recording. Experimental results using a real-world supply chain dataset demonstrate that the proposed framework achieves approximately 97% prediction accuracy with low error rates (MSE = 0.003 and RMSE = 0.02), outperforming several recent supply chain prediction models. The findings confirm the effectiveness of integrating GNN-based learning with blockchain-enabled verification for intelligent and scalable supply chain management.