Author(s)

Yuan Zhang^*, YanKun Li, Chao Si

Affiliation(s)

Electrical Department, China Institute of Atomic Energy, Beijing 102400, China.

Corresponding Author

 Yuan Zhang

ABSTRACT

The design of nuclear power plant electrical systems is characterized by an extensive and intricate body of standards, strong cross-disciplinary coupling, and stringent safety and compliance requirements. Traditional digital design workflows face significant limitations in knowledge reuse, intelligent decision support, and quality assurance. At the same time, general-purpose large language models (LLMs) exhibit hallucinations and lack reliability in safety-critical engineering domains. To address these challenges, this paper proposes an intelligent nuclear power electrical design methodology that integrates knowledge-graph-enhanced retrieval-augmented generation (Graph-RAG) with a multi-agent system (MAS).

First, a domain ontology covering IEC/GB standards, equipment parameter libraries, and design rules is constructed to provide a structured knowledge backbone. Second, a graph-guided hybrid retrieval strategy that combines semantic retrieval with graph path reasoning is designed to enhance retrieval accuracy and contextual relevance. Third, a multi-agent collaboration architecture—comprising requirement analysis, intelligent recommendation, automatic document generation, consistency checking, format review, and interactive question answering agents—is developed. Task orchestration is implemented through a directed acyclic graph (DAG), enabling parallel and coordinated execution of complex design workflows.

Theoretical analysis and scenario-based validation indicate that the proposed approach substantially reduces hallucination rates, improves design efficiency, and ensures the standard compliance of design outputs. Each design decision becomes traceable to explicit standard clauses and equipment parameters, demonstrating the feasibility of the method for the intelligent transformation of nuclear power electrical design and its potential transferability to other complex engineering systems.

KEYWORDS

Nuclear power electrical design; Knowledge graph; Retrieval-augmented generation; Multi-agent systems; Intelligent engineering

CITE THIS PAPER

Yuan Zhang, YanKun Li, Chao Si. Graph-RAG enhanced retrieval and multi-agent collaborative method for intelligent nuclear power electrical design. Journal of Computer Science and Electrical Engineering. 2025, 7(8): 1-14. DOI: https://doi.org/10.61784/jcsee3105.

REFERENCES

[1] Ji Z, Lee N, Frieske R, et al. Survey of hallucination in natural language generation. ACM Computing Surveys, 2023, 55(12): 1-38.

[2] Lewis P, Perez E, Piktus A, et al. Retrieval-augmented generation for knowledge-intensive NLP tasks. In: Proceedings of NeurIPS, 2020: 9459-9474.

[3] Gao Y, Xiong Y, Gao X, et al. Retrieval-augmented generation for large language models: A survey. arXiv preprint arXiv:2312.10997, 2023.

[4] Peng B, Galkin M, He P, et al. Graph retrieval-augmented generation: A survey. arXiv preprint arXiv:2408.08921, 2024.

[5] Wang Y, Zhong Q, Liu L, et al. A survey on LLM-based multi-agent systems: Workflow, infrastructure, and challenges. Autonomous Agents and Multi-Agent Systems, 2024, 38(2): 1-35.

[6] Buehler M J. Generative retrieval-augmented ontologic graph and multi agent strategies for interpretive large language model-based materials design. arXiv preprint arXiv:2310.19998, 2023.

[7] Huang Q, Peng S, Deng J, et al. A review of the application of artificial intelligence to nuclear reactors: Where we are and what’s next. Heliyon, 2023, 9(10): e13883.

[8] Guo, C, Yi, Y, Luo, W, et al. Design and prototype implementation of a nuclear power plant safety review decision-support system based on knowledge graphs. Nuclear Safety, 2024, 23(3): 55-62

[9] Yang R, Xue H, Razzak I, et al. Divide by question, conquer by agent: SPLIT RAG with question driven graph partitioning. arXiv preprint arXiv:2505.13994, 2025.

[10] Wang S, Fan W, Feng Y, et al. Knowledge graph retrieval-augmented generation for LLM based recommendation. arXiv preprint arXiv:2501.02226, 2025.

[11] Zhu X, Xie Y, Liu Y, et al.  Knowledge graph-guided retrieval augmented generation. In Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies. Albuquerque, New Mexico. Association for Computational Linguistics, 2025: 8912-8924. DOI: 10.18653/v1/2025.naacl-long.449.

[12] Shen H, Diao C, Vougiouklis P, et al. GeAR: Graph-enhanced agent for retrieval-augmented generation. arXiv preprint arXiv:2412.18431, 2024.