融合时空特征的光伏气象因子预测模型

Photovoltaic Meteorological Factor Prediction Model Fusing Spatial and Temporal Features

光伏发电极易受到天气的影响而具有波动性和不确定性,因此对气象因子的准确预测对光伏电站的运维具有重要意义。提出了一种基于深度学习的时空特征融合模型,实现对光伏气象因子的精准预测。在时间维度上,设计了一种改进的长短期记忆模块,融合注意力机制和遗传算法,得到最优注意力参数以提高预测精度;在空间维度上,将光伏电站所在区域按照经纬度划分,利用张量分解对区域内气象因子进行预测。在中国东南部某光伏系统的真实数据集上,对该模型的有效性进行了评估。结果表明,该模型在时间维度和空间维度均具有较高预测精度,同时对稀疏数据有较强的鲁棒性。

Photovoltaic power generation is volatile and uncertain due to its high susceptibility to weather,so accurate prediction of meteorological factors is important for the operation and maintenance of PV plants.A deep learning based spatial-temporal feature fusion prediction model is proposed to achieve accurate prediction of PV weather factors.In the temporal dimension,an improved long short-term memory network module is designed to integrate the attention mechanism and genetic algorithm to obtain the optimal attention parameters to improve the prediction accuracy;in the spatial dimension,the area where the PV plant is located is divided according to latitude and longitude,and the tensor decomposition is used to predict the meteorological factors in the area spatially.In the experiments,the effectiveness of the proposed model is evaluated using real datasets of PV plants and compared with the existing methods.The results show that the model has high prediction accuracy in both time and spatial dimensions,and has strong robustness to sparse data.