高电压技术2024，Vol.50Issue(3)：1025-1033,9.DOI:10.13336/j.1003-6520.hve.20230164
基于模型数据混合驱动的大规模双馈风电场并网数字孪生建模
Digital Twin Modeling for Grid-connected Large-scale Doubly Fed Wind Farms Based on Model-data Hybrid Drive
摘要
Abstract
Nowadays,the detailed model of doubly-fed wind turbine has multi-time scale dynamics and complex control links,which leads to difficulties,in modeling and that the detailed simulation modeling of large-scale doubly-fed wind turbines'farm requires huge computational forces.The grid-connected characteristics of actual wind turbines cannot be fed back into the simulation system in real time.The power output of wind farms is affected by many aspects,such as wind speed,wind direction and aging.The simulation system can not accurately fit the real power system.To solve this problem,this paper builds a hybrid model-data driven model for grid-connected power system of large-scale digital twin doubly-fed wind turbines'farm,which realizes real-time update of interactive feedback between wind turbine and power grid and reduces computational force requirements.At the same time,the virtual model matching real time with the wind turbine physical entity has higher output accuracy and meets the requirements of fine simulation.In this paper,each dou-bly-fed wind turbine uses the long and short term memory network(LSTM)to build a data-driven model.The wind farm architecture and the connected power grid select the common physical simulation model,and the digital twin technology is used to realize the two-way feedback of wind turbine physical model maintenance and virtual data model update.The example is tested by real measurement data,and the simulation results show that the digital twin model of the doubly-fed wind turbines'farm has better fitting ability and accuracy than the traditional model under different working conditions,and realizes real-time updating ability of the virtual model.This model is helpful to exploring the fault of a specific wind turbine,the change of control mode,the wind speed fluctuation and other conditions in large-scale wind farm grid-connection,and is helpful to studying the influence of wind turbine interaction and grid disturbance on wind farm.关键词
数字孪生/长短期记忆网络/大规模风电场并网/DFIG/实时更新/功率预测Key words
digital twin/LSTM/large-scale wind farm integration/DFIG/real-time update/power prediction引用本文复制引用
薛邵锴,秦文萍,张东霞,朱志龙,张永勤,李森良..基于模型数据混合驱动的大规模双馈风电场并网数字孪生建模[J].高电压技术,2024,50(3):1025-1033,9.基金项目
国家自然科学基金(51777132) （51777132）
山西省"1331"工程重点创新团队建设计划(1331KIRT). Project supported by National Natural Science Foundation of China(51777132),Shanxi Province"1331"Project Key Innovation Team Construction Program(1331KIRT). （1331KIRT）
