基于加权系数动态修正的短期风电功率组合预测方法  

Short-Term Wind Power Combination Forecasting Method Based on Dynamic Coefficient Updating

作　　者：王铮[1] Rui Pestana 冯双磊[1] 申洪[1] Luis Rosa 

机构地区：[1]中国电力科学研究院,北京市海淀区100192 [2]葡萄牙国家能源网公司

出　　处：《电网技术》2017年第2期500-507,共8页Power System Technology

基　　金：国家重点基础研究发展计划项目(973项目)(2016YFB0900502);国家自然科学基金项目(51477156);国家电网公司科技项目(NY7116021)~~

摘　　要：风电功率预测是解决大规模风电并网问题的有效手段之一,预测精度越高,越有利于提高电网运行的安全性和经济性。为了提高风电功率短期预测精度,通过组合预测方式弥补单一预测模型的局限性,针对各预测结果性能随预测时间尺度变化的问题,提出了不同时间断面差异化的组合预测方法,并根据风电功率的波动特性,恰当引入天气变化的持续信息,优化了15 min^4 h预测时间尺度下的预测精度。同时,针对各集合成员每日更新结果,通过在线建模方式,动态修正各集合单元组合权值,提高预测模型的适应性。实际算例表明,所提方法能够得到风电功率预测结果,组合系数变化性质符合实际原理,与目前常用的组合预测方法相比,预测精度更优。Wind power forecast is a way to solve large-scale wind power integration problems. Higher forecast accuracy is conducive to improve security and economy of power grid. Purpose of this paper is to improve short-term wind power forecast accuracy. Combination forecasting method was used to make up limitations caused by single forecasting model. Considering forecast performance change with prediction time-scale, this paper proposed differentiation combination forecast method according to different prediction time-scale. According to wind power fluctuation, persistence weather information was introduced, and forecast accuracy between 15min to 4h was improved, Finally, on-line model building method was used to optimize combination coefficients while forecast results were updated. Case study shows that this method can forecast wind power and the combination coefficients conform to actual theory. Compared with commonly used combination forecasting method, this method works with higher accuracy.

关 键 词：风力发电 风电功率预测 功率波动特性 集合预测 动态组合预测 

分 类 号：TM715[电气工程—电力系统及自动化]