摘要 对任意来流条件下的风电场发电功率进行准确预测,是提高电网对风电接纳能力的有效措施。针对大型风电场的功率预测采用单点位风速外推预测代表性差的局限,提出基于高斯混合模型(GMM)聚类的风电场短期功率预测方法。方法结合数据分布特征,利用GMM聚类将大型风电场划分为若干机组群,借助贝叶斯信息准则指标评价,获得风电场内最优机组分组方案。实际算例验证表明,按照小时级、月度级、年度级等时间尺度进行统计,所建立的GMM聚类模型均极大地提高了未分组的风电功率预测模型的准确性。相较于应用广泛的k-means聚类、层次凝聚聚类等方法,GMM聚类方法在分组功率预测中表现出了显著优势,为大型风电场短期功率预测模型的优化及运行经济性的提升提供了技术支持与依据。 Accurate forecasting of wind farm power under arbitrary inflow conditions is an effective way to improve the ability of the power grid to accept wind power.Aiming at the limitation of poor representativeness of single-point wind speed extrapolation for power forecasting of large-scale wind farms,a short-term power forecasting method of wind farm based on Gaussian mixture model(GMM)clustering is proposed.The method combines the characteristics of data distribution,uses GMM clustering to divide large-scale wind farms into several unit groups,and obtains the optimal unit grouping scheme in the wind farm based on the index evaluation of Bayesian information criterion.The actual example verifies that the established GMM clustering model greatly improves the accuracy of ungrouped wind power forecasting models based on hourly,monthly,annual and other time scales.Compared with widely used methods such as k-means clustering and hierarchical agglomerative clustering,the GMM clustering method shows significant advantages in grouped power forecasting,which provides technical support and basis for the optimization of short-term power forecasting models for large-scale wind farms and the improvement of the operation economy.
出处 《电力系统自动化》 EI CSCD 北大核心 2021年第7期37-43,共7页 Automation of Electric Power Systems
基金 国家重点研发计划资助项目(2016YFB0900500)。
关键词 风电机组 高斯混合模型聚类 合理性评价 功率预测 wind turbine Gaussian mixture model(GMM)clustering rationality evaluation power forecasting
