分布式光伏接入下配电网新能源消纳量自动预测研究
Research on Automatic Prediction of New Energy Consumption in Distribution Networks Under Distributed Photovoltaic Access
范宇航 1陈思旭 1李英楠 1王佳慧 2郑逸加3
作者信息
- 1. 国网天津市电力公司蓟州供电分公司,天津 301900
- 2. 国网陕西省电力有限公司经济技术研究院,陕西 西安 710065
- 3. 国网北京市电力公司,北京 100042
- 折叠
摘要
分布式光伏接入下配电网新能源消纳量预测难度提升,现行方法在实际应用中相对偏差较高,而且置信度比较低,无法达到预期的预测效果,为此提出分布式光伏接入下配电网新能源消纳量自动预测研究.从供电侧和用电侧2个方面选取新能源消纳的影响因素,并收集相关数据对其归一化处理,利用双向长短期记忆网络对消纳量影响因素数据学习分析,预测新能源消纳量,实现分布式光伏接入下配电网新能源消纳量自动预测.实验证明,设计方法预测相对偏差不超过1%,置信度不低于97%,可以实现对配电网新能源消纳量的精准自动化预测.
Abstract
The difficulty of predicting the consumption of new energy in the distribution network under distributed photovoltaic access has increased,and the current methods have relatively high deviations and low confidence in practical applications,which cannot achieve the expected prediction effect.Therefore,a study on automatic prediction of the consumption of new energy in the distribution network under distributed photovoltaic access is proposed.Select the factors affecting the consumption of new energy from two aspects:The power supply side and the power consumption side,and collect relevant data for normalization processing.Use a bidirectional long short-term memory network to learn and analyze the factors affecting the consumption amount,predict the consumption amount of new energy,and achieve automatic prediction of the consumption amount of new energy in the distribution grid under distributed photovoltaic access.Experimental results have shown that the design method predicts a relative deviation of no more than 1%and a confidence level of no less than 97%,which can achieve accurate and automated prediction of the consumption of new energy in the distribution network.
关键词
分布式光伏/配电网/新能源/消纳量/自动预测/双向长短期记忆网络Key words
distributed photovoltaic/distribution network/new energy/consumption/automatic prediction/bidirectional long short-term memory network引用本文复制引用
出版年
2024
NETL
NSTL
万方数据