基于网格化数值天气预报的区域光伏发电多输出功率预测方法
A Grid-based Numerical Weather Prediction Method for Multi-output Prediction of Regional Photovoltaic Power
战文华 1车建峰 2王勃 2丁禹2
作者信息
- 1. 国网内蒙古东部电力有限公司赤峰供电公司,内蒙古赤峰 024000
- 2. 可再生能源并网全国重点实验室(中国电力科学研究院有限公司),北京 100192
- 折叠
摘要
区域光伏的短期功率预测是省级及以上电网调控中心制定发电计划、提高光伏消纳率的重要基础之一.光伏短期功率预测本质上是构建数值天气预报与实际功率之间的映射模型,为了实现预测精度的提升,利用网格化的数值天气预报,采用残差网络建立区域光伏的多输出预测模型,充分挖掘区域光伏所属空间的气象资源分布与各光伏电站功率的关联关系,实现以网格化数值天气预报为输入的区域各光伏电站的功率预测.以实际运行数据进行仿真,结果表明,本文方法在各光伏电站的功率和总功率 2 个方面的预测结果均优于现有成熟方法.
Abstract
The short-term power prediction of regional photovoltaic(PV)is one of the important bases for the provincial and above power grid control center to make power generation plans and improve the photovoltaic consumption.In essence,short-term photovoltaic power prediction is to build a mapping model between numerical weather prediction and actual power.In order to improve the prediction accuracy,the grid-based numerical weather prediction and residual network were used in this paper to establish a multi-output prediction model of regional photovoltaic,and fully explore the correlation between the distribution of meteorological resources and the power of each photovoltaic power station in the regional photovoltaic space,thus identifying the grid-based numerical weather prediction as input and the power of each photovoltaic power station in the region as output.The simulation was carried out by use of actual operation data.The results showed that the proposed method was superior to the existing proven methods in predicting the power and total power of each photovoltaic power station.
关键词
光伏功率预测/网格化数值天气预报/残差网络/多输出模型Key words
photovoltaic power prediction/gird numerical weather prediction/residual network/multi-output forecast model引用本文复制引用
基金项目
国家电网内蒙古东部电力有限公司科技项目(SGMDCF00YWJS2000769)
出版年
2024
国家科技期刊平台
NSTL
万方数据