山西大学学报（自然科学版）2024，Vol.47Issue(6)：1201-1210,10.DOI:10.13451/j.sxu.ns.2024115
基于Informer的风电场超短期功率预测
Ultra-short-termd Power Prediction for Wind Farms Based on Informer
摘要
Abstract
Aiming at the problems of poor sustainability and stability of wind power in ultra-short-term wind electric power prediction,the varying degrees of influence of various factors in wind turbine output on ultra-short-term wind electric power prediction,and the limited attention mechanism and lack of global information of the original Transformer neural network architecture,a wind power prediction model based on the combination of Informer network architecture and Markov chain correction was proposed.The model uses algorithms including random forest and Lagrangian interpolation to preprocess data,uses the Informer neural network architecture to predict wind electric power,and uses Markov chains to correct errors in ultra-short-term wind power prediction results to improve model prediction accuracy.Field experiment results show that the model improves the prediction time step and prediction accuracy of wind electric power,with the monthly average accuracy reaching up to 97%and the annual average accuracy reaching 95%.关键词
超短期风功率预测/Informer/多头稀疏概率自注意力机制/预测误差修正Key words
ultra-short term wind power forecast/Informer/multi-headed self-attention/prediction error correction分类
数理科学引用本文复制引用
李超峰,原升耀,王灵梅,王贺贤,贾成真,刘玉山,常馨元..基于Informer的风电场超短期功率预测[J].山西大学学报（自然科学版）,2024,47(6):1201-1210,10.基金项目
山西省重点研发计划项目(202202060301004) （202202060301004）
