基于特征挖掘的GRU-A光伏发电功率预测  

Photovoltaic Power Prediction Based on Feature Mining and GRU-A

作　　者：张进[1] 刘运[1] 彭曙蓉[2] ZHANG Jin;LIU Yun;PENG Shurong(College of Electrical&Control Engineering,Shanxi University of Science&Technology,Xi’an 710021,China;College of Electrical&Information Engineering,Changsha University of Science&Technology,Changsha 410114,China)

机构地区：[1]陕西科技大学电气与控制工程学院,西安710021 [2]长沙理工大学电气与信息工程学院,长沙410114

出　　处：《实验室研究与探索》2020年第5期25-30,49,共7页Research and Exploration In Laboratory

基　　金：国家自然科学基金项目(51272148);湖南省教育厅项目创新平台开放基金(17K001)。

摘　　要：提出了一种基于特征挖掘的GRU-A光伏发电功率预测模型,可有效、精确地预测短期光伏发电功率。对光伏发电功率的影响因素进行深度挖掘并构造出相应特征群,搭建GRU-A模型作为光伏发电功率预测模型,并以构造的特征群作为输入,来实现光伏发电功率预测。利用某地的实际光伏数据,通过仿真实验证明基于特征挖掘的GRU-A模型相比于其他模型有更好的预测精度及稳定性。In this paper,a GRU-A photovoltaic power generation prediction model based on feature mining is proposed.It can effectively and accurately predict short-term photovoltaic power generation.Firstly,the influencing factors of photovoltaic power generation are deeply excavated and the corresponding feature groups are constructed.Then a GRU-A model is built as the photovoltaic power prediction model,and the constructed feature group is used as input to realize the prediction of photovoltaic power generation.Finally,using the actual photovoltaic data of a certain place,the simulation results show that the prediction accuracy and stability of the GRU-A model based on feature mining are significantly superior to other models.

关 键 词：特征挖掘 门控循环模型 注意力机制 光伏发电功率预测 

分 类 号：TM615[电气工程—电力系统及自动化]