基于EEMD-ANN的自适应光伏日电量预测方法  

Adaptive Photovoltaic Daily Power Forecasting Algorithm based on EEMD-ANN

作　　者：黄娟 赵鹏 王聚博 凌宇龙 苏洋 赵闻音 HUANG Juan;ZHAO Peng;WANG Ju-bo;LING Yu-long;SU Yang;ZHAO Wen-yin(SPIC Jiangxi Electric Power Co.,Ltd.,Nanchang 330000,China;SPIC Central Research Institute,Beijing 102200,China)

机构地区：[1]国家电投集团江西电力有限公司,江西南昌330000 [2]国家电投集团科学技术研究院有限公司,北京102200

出　　处：《节能技术》2024年第5期418-424,共7页Energy Conservation Technology

摘　　要：随着清洁能源的持续发展,我国光伏电源装机规模不断增大。为了应对其随机性、波动性、不确定性等特点给电网安全运行带来的严峻挑战,研究中结合集合经验模态分解(EEMD)方法对原始时间序列进行处理,将其分解为有限且少量的振荡模式,形成更清晰的信号输入,再通过人工神经网络(ANN)实现历史数据的规律挖掘,构建了基于EEMD-ANN的自适应光伏日电量预测模型。以我国南方某光伏电站日电量过程为例的结果表明,该模型获得的预测结果具有较好的预测精度,是一种实用性较强的光伏电量预测方法。With the continuous development of clean energy,the installed capacity of photovoltaic power sources in China is constantly increasing.In order to cope with the severe challenges brought by its randomness,volatility,uncertainty and other characteristics to the safe operation of the power grid,the research combines the set empirical mode decomposition(EEMD)method to process the original time series,decomposing it into limited and small oscillation modes to form clearer signal inputs.Then,artificial neural networks(ANN)are used to mine the patterns of historical data,and an adaptive photovoltaic daily electricity prediction model based on EEMD-ANN is constructed.The results of taking the daily electricity consumption process of a photovoltaic power station in southern China as an example show that the prediction results obtained by the model have good prediction accuracy and are a practical method for predicting photovoltaic electricity consumption.

关 键 词：光伏发电 集合经验模态分解(EEMD) 人工神经网络(ANN) 非平稳 自适应预测 

分 类 号：TM615[电气工程—电力系统及自动化]