ZHILIU: AI-CFD multi-agent accelerated simulation platform for engineering applications

Liu Cheng; Yang Siyao; Wei Xinnan; Chen Yixiao; Lin Kun

doi:10.7638/kqdlxxb-2025.0220

English Version Journal Login

您当前的位置：

首页 >

文章列表页 >

ZHILIU: AI-CFD multi-agent accelerated simulation platform for engineering applications

更新时间：2026-05-28

- ZHILIU: AI-CFD multi-agent accelerated simulation platform for engineering applications
- Vol. 44, Issue 3, Pages: 123-131(2026)
- 作者机构：
  
  哈尔滨工业大学(深圳) 智能土木与海洋工程学院,深圳,518055
- 作者简介：
- 基金信息：
- DOI：10.7638/kqdlxxb-2025.0220
  CLC： V211.3
- Published：2026
- 稿件说明：
移动端阅览
刘承, 杨思尧, 魏昕楠, 等. 智流：面向工程场景的AI-CFD多智能体加速仿真平台[J]. 空气动力学学报, 2026, 44(3): 123−131.

Liu Cheng, Yang Siyao, Wei Xinnan, et al. ZHILIU: AI-CFD multi-agent accelerated simulation platform for engineering applications[J]. 2026, 44(3): 123-131.
刘承, 杨思尧, 魏昕楠, 等. 智流：面向工程场景的AI-CFD多智能体加速仿真平台[J]. 空气动力学学报, 2026, 44(3): 123−131. DOI： 10.7638/kqdlxxb-2025.0220

Liu Cheng, Yang Siyao, Wei Xinnan, et al. ZHILIU: AI-CFD multi-agent accelerated simulation platform for engineering applications[J]. 2026, 44(3): 123-131. DOI： 10.7638/kqdlxxb-2025.0220.

摘要

针对工程CFD仿真存在门槛高、流程配置复杂、经验依赖强以及收敛与调参成本高等关键难题，本文构建了“智流”——一个可追踪、可复用且可本地化的AI-CFD多智能体加速仿真平台。该平台采用以大语言模型为中枢的多阶段编排，融合检索增强生成与工程知识图谱，将自然语言需求转化为结构化任务；在多智能体协同框架下实现异常诊断与自愈调参，统一封装几何、网格、求解器以及后处理接口，并提供本地部署、权限控制与日志审计的合规保障，从而实现从需求解析、建模、求解、到可视化报告的端到端自动化。平台具备自动识别网格或收敛异常并可通过参数优化恢复计算的能力，支持全阶段人工介入与人机协同，在保障求解稳定性的同时满足多样性的仿真需求；此外，该平台支持并发计算与GPU加速，有效提升任务吞吐与仿真效率。通过对标准算例开展“智流”平台与常规仿真方法的计算效率对比研究，结果表明：“智流”平台能显著减少人工介入时间和操作步数，其中人工介入时间减少了65.26%，总操作步数仅为常规仿真的1/8；并在求解稳定性方面表现一定的优势。本研究主要基于国产开源CFD平台——风雷软件（NNW-PHengLEI），并调用其核心求解器模块，为CFD从工具驱动向智能协同转型提供可落地路径，可适配科研与工业场景中的规范化、合规化仿真生产，在保证仿真精度的同时显著提升仿真效率。

Abstract

To address the longstanding challenges in engineering CFD—namely the high entry barrier

complex workflow configuration

heavy dependence on expert experience

and the high cost of convergence tuning—this work develops ZHILIU

a traceable

reusable

and fully localizable AI-CFD multi-agent platform for accelerated simulation. The platform adopts a multi-stage orchestration centered on a large language model and integrates retrieval-augmented generation (RAG) with an engineering knowledge graph to transform natural-language requirements into structured tasks. Within a coordinated multi-agent framework

it performs automated anomaly diagnosis and self-healing parameter tuning

while providing unified interfaces for geometry preparation

meshing

solvers

and post-processing. Local deployment

permission management

and audit logging further ensure security and compliance. The system enables end-to-end automation from requirement parsing and model setup to solving and visualization reporting. Mesh or convergence anomalies can be automatically detected and corrected through parameter optimization

while human intervention can be incorporated at any stage to support flexible human-AI collaboration and diverse simulation needs. The platform also supports concurrent execution and GPU acceleration to increase throughput and substantially enhance overall computational efficiency. Standard benchmark comparisons between the ZJILIU platform and conventional CFD workflows were conducted. The results show that ZHILIU significantly reduces manual intervention by 65.26% and cuts operational steps to just 1/8 of the conventional approach. It also demonstrates an advantage in solution stability. Overall

ZHILIU provides a practical and deployable pathway for transitioning CFD from tool-centric workflows to intelligent

collaborative simulation. This study is primarily based on the domestic open-source CFD platform (NNW-PHengLEI)

and invokes its core solver module. It provides a feasible path for transforming CFD from tool-driven to intelligent collaborative approaches. The platform improves efficiency without sacrificing accuracy

and is suitable for both research and industrial applications.

关键词

Keywords

references

Wijesooriya K, Mohotti D, Lee C K, et al. A technical review of computational fluid dynamics (CFD) applications on wind design of tall buildings and structures: Past, present and future[J]. Journal of Building Engineering, 2023, 74: 106828. doi: 10.1016/j.jobe.2023.106828

Zawawi M H, Saleha A, Salwa A, et al. A review: Fundamentals of computational fluid dynamics (CFD)[C]//Proc of the AIP Conference Proceedings, Ho Chi Minh, Vietnam. Author(s), 2018: 020252. doi: 10.1063/1.5066893

Wang F Z, Animasaun I L, Muhammad T, et al. Recent advancements in fluid dynamics: drag reduction, lift generation, computational fluid dynamics, turbulence modelling, and multiphase flow[J]. Arabian Journal for Science and Engineering, 2024, 49(8): 10237−10249. doi: 10.1007/s13369-024-08945-3

Kodman J B, Singh B, Murugaiah M. A comprehensive survey of open-source tools for computational fluid dynamics analyses[J]. Journal of Advanced Research in Fluid Mechanics and Thermal Sciences, 2024, 119(2): 123−148. doi: 10.37934/arfmts.119.2.123148

Sun X X, Cao W B, Shan X L, et al. A generalized framework for integrating machine learning into computational fluid dynamics[J]. Journal of Computational Science, 2024, 82: 102404. doi: 10.1016/j.jocs.2024.102404

Rahman M S, Hazra S, Chowdhury I A. Advancing computational fluid dynamics through machine learning: A review of data-driven innovations and applications[J]. Journal of Fluid Mechanics and Mechanical Design, 2024, 6(2): 42−51. doi: 10.46610/jfmmd.2024.v06i02.005

Du B C, Shen E N, Wu J P, et al. Aerodynamic prediction and design optimization using multi-fidelity deep neural network[J]. Aerospace, 2025, 12(4): 292. doi: 10.3390/aerospace12040292

Ponzini R, Salvadore F, Begovic E, et al. Automatic CFD analysis of planing hulls by means of a new web-based application: Usage, experimental data comparison and opportunities[J]. Ocean Engineering, 2020, 210: 107387. doi: 10.1016/j.oceaneng.2020.107387

Liu Z J. Fully automated CFD simulation system research based on design scheme tree[J]. Scientific Reports, 2025, 15: 3975. doi: 10.1038/s41598-024-83582-2

Mao R, Lan Y E, Liang L F, et al. Rapid CFD prediction based on machine learning surrogate model in built environment: A review[J]. Fluids, 2025, 10(8): 193. doi: 10.3390/fluids10080193

Caccavaro D, Tagliafierro B, Bilotta G, et al. An analysis and comparison of the hydrodynamic behavior of ships using mesh-based and meshless computational fluid dynamics simulations[J]. Fluids, 2024, 9(11): 266. doi: 10.3390/fluids9110266

Tang Zhigong, Qian Weiqi, He Lei, et al. Thoughts and prospects on large model research in aerodynamics[J]. Acta Aerodynamica Sinica, 2024, 42(12): 1−11 (in Chinese). doi: 10.7638/kqdlxxb-2024.0056

Tang Zhigong, Zhu Linyang, Xiang Xinghao , et al. Some research progress and prospect of Intelligent Aerodynamics[J]. Acta Aerodynamica Sinica, 2023, 41(7): 1−35 (in Chinese). doi: 10.7638/kqdlxxb-2023.0128

Vardhan H, Timalsina U, Sandborn M, et al. Anvil: an integration of artificial intelligence, sampling techniques, and a combined CAD-CFD tool[C]// 2024 IEEE Workshop on Design Automation for CPS and IoT (DESTION), Hong Kong, China. IEEE, 2024: 16-21. doi: 10.1109/DESTION62938.2024.00009

Chen T L, Li R B, Hu X S, et al. Machine learning as CFD surrogate models for rapid prediction of building-related physical fields: A review of methods and state-of-the-art[J]. Building and Environment, 2025, 285: 113667. doi: 10.1016/j.buildenv.2025.113667

Zhao L G, Zhou Q, Li M Y, et al. Evaluating different CFD surrogate modelling approaches for fast and accurate indoor environment simulation[J]. Journal of Building Engineering, 2024, 95: 110221. doi: 10.1016/j.jobe.2024.110221

Markarian E, Qiblawi S, Krishnan S, et al. Informing building retrofits at low computational costs: A multi-objective optimisation using machine learning surrogates of building performance simulation models[J]. Journal of Building Performance Simulation, 2024: 1−17. doi: 10.1080/19401493.2024.2384487

Views

下载量

CSCD

Alert me when the article has been cited

提交

Tools

Publicity Resources

Development and open source application progress of NNW-PHengLEI

Dynamic ablation morphology measurement of thermal protection materials using structured light vision

Multi-factor analysis and simulation of particle erosion characteristics of fan blades

An Ac-criteria-based optimal input design method

Optimization of hot-air anti-icing structures based on neural networks and genetic algorithms

Related Author

Zhao Zhong

He Xianyao

He Lei

Huang Can

Related Institution

Computational Aerodynamic Institute of China Aerodynamics Research and Development Center, Mianyang

Sichuan Tianfu Fluid Big Data Research Center, Chengdu

AI问答

⁰