首页|期刊导航|应用数学和力学|覆冰分裂导线气动系数多保真代理模型

覆冰分裂导线气动系数多保真代理模型OA

A Multi-Fidelity Surrogate Model for Aerodynamic Coefficients of Iced Bundle Conductors

中文摘要英文摘要

针对新月形覆冰四分裂、六分裂和八分裂导线,采用计算流体力学方法模拟覆冰导线绕流场,获得了不同冰风条件下覆冰子导线阻力、升力和扭矩系数随风攻角的变化规律.利用风洞试验和数值模拟得到的气动系数高、低保真数据,结合 BP 神经网络、随机森林算法和特征增强技术,建立了以子导线位置、冰厚、凝冰角、风速、湍流度、风攻角为输入,覆冰子导线气动系数为输出的多保真代理模型.通过 Den Hartog 和 Nigol 系数的气动稳定性区域分析和典型覆冰分裂导线舞动模拟,验证了代理模型的有效性.基于多保真代理模型设计了以导线分裂数、冰厚、凝冰角、风速、湍流度为输入,各子导线气动系数随风攻角变化曲线为输出的软件,为覆冰线路舞动及其防治技术研究奠定了基础.

For quad-,six-,and eight-bundle conductors with crescent-shaped ice,the air flow fields around the iced conductors under different icing and wind conditions were numerically simulated with the computation-al fluid dynamics,and the variations of drag,lift,and torque coefficients of each iced sub-conductor vs.the wind attack angle were obtained.With the high-fidelity data of aerodynamic coefficients measured by wind tun-nel tests and the low-fidelity data by numerical simulations,a multi-fidelity surrogate model was built by means of the BP neural networks,the random forest algorithm and the feature enhancement technique,with the sub-conductor position,the ice thickness,the ice accretion angle,the wind speed,the turbulence intensity and the wind attack angle as the input,and the aerodynamic coefficients of each sub-conductor as the output.The effi-ciency of the surrogate model was verified through the analysis of aerodynamic stability regions based on the Den Hartog and Nigol coefficients,and through the galloping simulations of typical iced bundle conductors.By this multi-fidelity surrogate model,a software with the bundle number of conductor lines,the ice thickness,the ice accretion angle,the wind speed and the turbulence intensity as the input,and the aerodynamic coeffi-cient vs.the wind attack angle for each iced sub-conductor as the output,was developed.This work lays a foundation for the investigation of galloping of iced transmission lines and its prevention technologies.

杨辉;严波;高英博;张博;吕中宾;刘光辉;伍兴旺;邓华超

重庆大学 航空航天学院,重庆 400044重庆大学 航空航天学院,重庆 400044重庆大学 航空航天学院,重庆 400044国网河南省电力公司电力科学研究院,郑州 450052国网河南省电力公司电力科学研究院,郑州 450052国网河南省电力公司电力科学研究院,郑州 450052重庆大学 航空航天学院,重庆 400044重庆大学 航空航天学院,重庆 400044

数理科学

覆冰分裂导线气动系数机器学习多保真代理模型

iced bundled conductoraerodynamic coefficientmachine learningmulti-fidelity surrogate model

《应用数学和力学》 2026 (9)

1159-1178,20

国家电网公司科学技术项目(5500-202324543A-3-2-ZN)

10.21656/1000-0887.460131

评论