Yan Xiong
Yan Xiong has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Dazhi Zhang
China Nuclear Power Operation Technology, Co., Ltd., Wuhan 430223, China
1 shared publication - Huabing Zhou
Hubei Key Laboratory of Intelligent Robot, Wuhan Institute of Technology, Wuhan 430205, China
1 shared publication - Jing-Chen Lin
Department of Thermal Science and Energy Engineering, University of Science and Technology of China, Hefei 230026, China
1 shared publication - Jun Luo
China Nuclear Power Operation Technology Corporation, LTD., Wuhan 430223, China
1 shared publication - Luo Jun
Department of Engineering Physics, Tsinghua University, Beijing 100084, China
1 shared publication - Xingyan Lu
China Nuclear Power Operation Technology Corporation, LTD., Wuhan 430223, China
1 shared publication
Latest Publications
- Open Access
Intelligent Parameter Tuning Strategy for Nuclear Power Simulation Systems Based on Soft Actor–Critic
The effective tuning of extensive, high-dimensional parameter sets within nuclear power simulation systems presents considerable challenges for conventional, experience-based methodologies. Such approaches often exhibit limitations including slow convergence rates, susceptibility to local optima, and inadequate adaptability across diverse scenarios. This study introduces an automated parameter tuning strategy employing the Soft Actor–Critic (SAC) reinforcement learning approach. A SAC-based agent is formulated, where state representations, action spaces, and reward functions are specifically defined to tune parameters under predefined constraints. This facilitates the automated adjustment of complex parameter configurations. Comparative experimental evaluations indicate that the proposed SAC method reaches a solution quality comparable to that of the established baseline techniques while producing markedly more consistent results across independent runs, offering a reliable and repeatable methodology for parameter tuning in this domain.
Open AccessIntelligent Parameter Tuning Strategy for Nuclear Power Simulation Systems Based on Soft Actor–Critic
The effective tuning of extensive, high-dimensional parameter sets within nuclear power simulation systems presents considerable challenges for conventional, experience-based methodologies. Such approaches often exhibit limitations including slow convergence rates, susceptibility to local optima, and inadequate adaptability across diverse scenarios. This study introduces an automated parameter tuning strategy employing the Soft Actor–Critic (SAC) reinforcement learning approach. A SAC-based agent is formulated, where state representations, action spaces, and reward functions are specifically defined to tune parameters under predefined constraints. This facilitates the automated adjustment of complex parameter configurations. Comparative experimental evaluations indicate that the proposed SAC method reaches a solution quality comparable to that of the established baseline techniques while producing markedly more consistent results across independent runs, offering a reliable and repeatable methodology for parameter tuning in this domain.
- Open Access
- 11 June 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (12) , 5885
Reinforcement Learning-Enhanced Large Language Models for Automated Modeling of Nuclear Thermal-Hydraulic Systems: A Plan-and-Act Agent Framework
Automating system-level nuclear thermal-hydraulic (T-H) model construction remains challenging because platform-specific API syntax, graph connectivity, parameter dependency ordering, and solver admissibility must be satisfied simultaneously. This study develops a closed-loop modeling framework on the SAFRI platform by combining supervised fine-tuning (SFT), a Plan-and-Act agent with retrieval-grounded parameter completion, and reinforcement learning based on group relative policy optimization (GRPO). The SFT stage uses a 6003-record domain corpus derived from expert-authored or expert-verified SAFRI modeling exemplars, while system-level generalization is evaluated on a held-out 50-case in-house evaluation set separated at the case-template level. At the component level, LoRA-adapted Qwen3-8B achieves 100% code accuracy, compared with 50% for zero-shot and 74% for one-shot prompting. At the system level, the SFT agent attains a 100% syntax success rate (SSR), 90% topology success rate (TSR), and 72.4% physical convergence rate (PCR), showing that local API correctness is insufficient for solver-valid model assembly. After GRPO training with schema, topology, physics, and sequence rewards, the full SAFRI-SFT-RL agent reaches a 100% SSR, 100% TSR, and 88.8% PCR on the in-house evaluation set, while an error self-healing loop resolves execution-time failures in an average of 2.3 corrective iterations. These results show that solver-grounded reinforcement learning is effective for closing the gap between syntactically correct script generation and physically convergent nuclear T-H model construction.
- 11 June 2026
- Published by MDPI AG in Applied Sciences
- Vol. 16 (12) , 5885
Open AccessReinforcement Learning-Enhanced Large Language Models for Automated Modeling of Nuclear Thermal-Hydraulic Systems: A Plan-and-Act Agent Framework
Automating system-level nuclear thermal-hydraulic (T-H) model construction remains challenging because platform-specific API syntax, graph connectivity, parameter dependency ordering, and solver admissibility must be satisfied simultaneously. This study develops a closed-loop modeling framework on the SAFRI platform by combining supervised fine-tuning (SFT), a Plan-and-Act agent with retrieval-grounded parameter completion, and reinforcement learning based on group relative policy optimization (GRPO). The SFT stage uses a 6003-record domain corpus derived from expert-authored or expert-verified SAFRI modeling exemplars, while system-level generalization is evaluated on a held-out 50-case in-house evaluation set separated at the case-template level. At the component level, LoRA-adapted Qwen3-8B achieves 100% code accuracy, compared with 50% for zero-shot and 74% for one-shot prompting. At the system level, the SFT agent attains a 100% syntax success rate (SSR), 90% topology success rate (TSR), and 72.4% physical convergence rate (PCR), showing that local API correctness is insufficient for solver-valid model assembly. After GRPO training with schema, topology, physics, and sequence rewards, the full SAFRI-SFT-RL agent reaches a 100% SSR, 100% TSR, and 88.8% PCR on the in-house evaluation set, while an error self-healing loop resolves execution-time failures in an average of 2.3 corrective iterations. These results show that solver-grounded reinforcement learning is effective for closing the gap between syntactically correct script generation and physically convergent nuclear T-H model construction.
- 13 January 2026
- Published by Springer Nature
- p. 50-64
Connectivity Relationship Recognition in Piping and Instrumentation Diagrams Using Graph Neural Networks
- 13 January 2026
- Published by Springer Nature
- p. 50-64
Connectivity Relationship Recognition in Piping and Instrumentation Diagrams Using Graph Neural Networks