中国电机工程学报2017，Vol.37Issue(18)：5238-5247,10.DOI:10.13334/j.0258-8013.pcsee.161368
基于因子分析和神经网络分位数回归的月度风电功率曲线概率预测
Month-ahead Wind Power Curve Probabilistic Prediction Based on Factor Analysis and Quantile Regression Neural Network
摘要
Abstract
Concerning the existing problems such as large number of highly correlated variables,few available weather information and high uncertainty in month-ahead wind power curve prediction,a new month-ahead wind power curve probabilistic prediction method based on factor analysis and quantile regression neural network (QRNN) was proposed.By means of the factor analysis method to reduce the dimensionality of intraday hourly wind power time series vector,the independent common factors were extracted and used as the predictor variables to separately build QRNN probabilistic forecasting models with the daily weather features as inputs.Based on these probabilistic forecasting models,the probability density functions of wind power common factors of next 30 days were predicted by inputting the daily weather features from medium-range weather forecasts.By simulating common factors and specific factors follow the predicted distributions to recover wind power forecast curves day by day,statistical scenarios of wind power production in next month were generated finally.The accuracy,adaptability and high efficiency of the proposed method have been verified by the forecasting results of two actual wind farms,which can provide a new feasible solution for the probabilistic medium-and long-term wind power forecasting.关键词
月度风电功率预测/因子分析/神经网络分位数回归/中期天气预报/概率预测Key words
month-ahead wind power prediction/factor analysis/quantile regression neural network/medium-range weather forecasting/probabilistic prediction分类
信息技术与安全科学引用本文复制引用
李丹,任洲洋,颜伟,朱继忠,赵霞,余娟..基于因子分析和神经网络分位数回归的月度风电功率曲线概率预测[J].中国电机工程学报,2017,37(18):5238-5247,10.基金项目
国家自然科学基金项目(51177178,51677012) （51177178,51677012）
重庆市科委基础与前沿研究计划项目(cstc2013jcyjA90001).The National Natural Science Foundation of China (51177178 & 51677012) （cstc2013jcyjA90001）
The Foundation and Frontier Research Project of Chongqing Science and Technology Commission (cstc2013jcyjA90001). （cstc2013jcyjA90001）
