摘要
为准确预测光伏电站的发电功率,帮助电网调度部门制定合理的调度计划,文章提出一种基于SGMD(Symplectic Geometry Mode Decomposition)、CNN(Convolutional Neural Networks)和BiGRU(Bidirectional Gate Recurrent Unit)的光伏发电功率预测模型。首先,利用辛几何分解将历史功率分解为不同模态;其次,结合天气数据输入CNN-BiGRU组合模型进行预测;最后,将预测结果整合。该模型选用新疆某光伏电站2019年运行数据分别在短期与中长期不同预测范围内进行预测实验,实验结果表明,此模型的通用性和辛几何分解算法在提高光伏功率预测精度上具有一定的研究价值。
To accurately predict the power generation of photovoltaic power plants and assist the power grid scheduling department in making scheduling plans.Propose a photovoltaic power generation power prediction model based on symbolic geometry mode decomposition,rotational neural networks,and bidirectional gate current unit.Firstly,the historical power is decomposed into different modes using symplectic geometric decomposition,and then combined with weather data input into the CNN-BiGRU combination model for prediction.Finally,the prediction results are integrated.The 2019 operating data of a photovoltaic power station in Xinjiang was selected for prediction experiments and comparison experiments in different short-term and medium to long-term prediction ranges,verifying the universality of this model and the good research value of the symplectic geometric decomposition algorithm in improving the accuracy of photovoltaic power prediction.
作者
邹邦杰
刘国巍
Zou Bangjie;Liu Guowei(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan 232001,China)
出处
《无线互联科技》
2023年第23期128-130,共3页
Wireless Internet Technology
关键词
光伏功率预测
辛几何模态分解
卷积神经网络
双向门控单元
photovoltaic power forcasting
symplectic geometry mode decomposition
convolutional neural networks
bidirectional gate recurrent unit
作者简介
邹邦杰(1999-),男,安徽六安人,硕士研究生,研究方向:电力电子与新能源。
