引用本文: | 葛磊蛟,秦羽飞,刘嘉恒,白星振.基于相似日与BA-WNN的分布式光伏数据虚拟采集方法[J].电力自动化设备,2021,41(6): |
| GE Leijiao,QIN Yufei,LIU Jiaheng,BAI Xingzhen.Virtual acquisition method of distributed photovoltaic data based on similarity day and BA-WNN[J].Electric Power Automation Equipment,2021,41(6): |
|
摘要: |
我国分布式光伏电站具有点多面广、分散无序等特点,这使得光伏运维数据采集面临投资成本巨大且实时采集任务繁重等难题。提出一种基于相似日、蝙蝠算法与小波神经网络(BA-WNN)相结合的分布式光伏数据虚拟采集方法,实现在同一区域内仅1座分布式光伏电站安装完备的数据采集装置、其他分布式光伏电站安装价格较低的电流采集终端的场景下,完成区域范围内所有电站运维数据的虚拟采集。利用灰色关联度与余弦相似度构成相似性综合指标进行相似日的选取并建立相似日样本集;利用蝙蝠算法对小波神经网络的权值、伸缩因子与平移因子进行不断调整优化,并利用训练好的BA-WNN模型进行区域范围内分布式光伏电站出力数据的“实时+虚拟”采集。以区域范围内9座分布式光伏电站为例,验证了所提方法的可行性与有效性。 |
关键词: 分布式光伏 相似日 相似性综合指标 蝙蝠算法 小波神经网络 虚拟采集 |
DOI:10.16081/j.epae.202106011 |
分类号:TM615 |
基金项目:国家重点研发计划项目(2018YFB1500800);国家自然科学基金资助项目(51807134);国家电网有限公司总部科技项目(高渗透率分布式光伏接入电网动态特性及稳定运行控制技术研究) |
|
Virtual acquisition method of distributed photovoltaic data based on similarity day and BA-WNN |
GE Leijiao1, QIN Yufei2, LIU Jiaheng1, BAI Xingzhen2
|
1.Key Laboratory of Smart Grid of Ministry of Education, Tianjin University, Tianjin 300072, China;2.School of Electrical and Automation Engineering, Shandong University of Science and Technology, Qingdao 266590, China
|
Abstract: |
The distributed photovoltaic power stations in China have the characteristics of widely and disorderly distributed, etc, which makes photovoltaic operation and maintenance data acquisition face problems of huge investment cost and heavy task of real-time acquisition. A virtual acquisition method of distributed photovoltaic data is proposed based on the combination of similar day and BA-WNN(Bat Algorithm and Wavelet Neural Network),which realizes virtual acquisition of operation and maintenance data of all power stations in regional range under the scenario that only one distributed photovoltaic power station is equipped with complete data acquisition device while other distributed photovoltaic power stations are equipped with low-priced current acquisition terminals in a region. The gray correlation degree and cosine similarity are used to form a similarity composite index for selecting similar days and a sample set of similar days is established. The bat algorithm is used to continuously adjust and optimize the weights, expansion factors and translation factors of wavelet neural network, and the trained BA-WNN model is used for “real time + virtual” acquisition of the output data of distributed photovoltaic power stations in regional range. Nine distributed photovoltaic power stations in regional range are taken as examples to verify the feasibility and effectiveness of the proposed method. |
Key words: distributed photovoltaic similar day similarity composite index bat algorithm wavelet neural network virtual acquisition |