SC-DiET: Smart Contract Vulnerability Detection Based on Semantic Extraction and Graph Fusion for Distributed Energy Trading

The growing proliferation of distributed renewable energy resources and edge computing infrastructures is driving a paradigm shift toward decentralized peer-to-peer energy trading for energy systems. It enables prosumers to exchange electricity directly without relying on centralized utilities. Blockchain-based smart contracts are increasingly deployed to automate transaction execution and ensure trust among system participants. However, the inherent complexity of these contracts may introduce security vulnerabilities that can threaten the integrity, fairness, and financial safety of energy trading processes. This paper presents a vulnerability detection method specifically designed for blockchain-based energy trading environments, combining semantic code analysis with structural modeling of contract behavior. By capturing both logical and interaction-level features within energy-related contracts, the proposed method effectively detects potential vulnerabilities that may otherwise be overlooked. Experimental results on real-world datasets demonstrate that the proposed method achieves superior accuracy, precision, and robustness, with an average F1-score of 87.14%. Moreover, it shows strong capability in identifying complex logical flaws in practical contract scenarios. These results highlight the method’s potential to enhance the security, reliability, and trustworthiness of decentralized energy trading systems in increasingly automated and prosumer-driven energy networks.