摘要
针对风电场风速预测中单一数据源导致的精度问题,提出基于多源数据融合的风速快速预测方法。通过相空间重构优化数据表示,设计门控卷积网络提取异常特征,并引入风机嵌入技术和时序记忆网络捕捉关键信息。实验显示,该方法在迭代200次时适应度函数值显著降低,对42台风电机组的风速预测偏差控制在0.05m/s内,预测精度显著提升。
This study proposes a fast wind speed prediction method based on multi-source data fusion to address the accuracy issue caused by a single data source in wind farm wind speed prediction.This study optimizes data representation through phase space reconstruction,designs a gated convolutional network to extract anomalous features,and introduces wind turbine embedding technology and temporal memory network to capture key information.The experiment showed that the fitness function value of this method significantly decreased after 200 iterations,and the wind speed prediction deviation for 42 wind turbines was controlled within 0.05 m/s,resulting in a significant improvement in prediction accuracy.
作者
万伟
李增军
陈旭
张辉
魏秀章
WAN Wei;LI Zengjun;CHEN Xu;ZHANG Hui;WEI Xiuzhang(Julu County Construction Investment Wind Energy Co.,Ltd.,Xingtai,Hebei 055250,China)
出处
《自动化应用》
2025年第16期140-142,共3页
Automation Application
关键词
多源数据融合
时间序列
风电场
风速预测
相空间重构
multi-source data fusion
time series
wind farm
wind speed prediction
phase space reconstruction
作者简介
万伟,男,1973年生,高级工程师,研究方向为风力发电场建设及运维管理。
