摘要
随着光伏发电大规模并入电网,由此产生的源端可供电量不确定性问题日益突出,精准预测光伏出力,对电网资源优化配置、提升光伏消纳能力起着关键的支撑作用。通过研究影响光伏出力的关键要素,以历史气象特征数据为输入构建光伏出力预测模型,在实践过程中,对存在的主客观问题,从算法层面进行了模型优化。(1)针对光伏电站历史运行样本量小、气象特征变化多,导致诸多稀疏特征样本的问题,引入了ADASYN自适应采用算法,进行数据集重平衡;(2)通过XGBoost算法搭建了基于气象特征的光伏出力模型,并与传统的BP神经网络进行比较。通过某光伏电站的实际历史数据预测结果比较,结合ADASYN过采样和XGBoost算法,能有效提升模型的准确性;较BP神经网络相比,ADASYN-XGBoost算法的MAE、RMSE、MAPE和R2分别提高了66.7%、68.9%、58.0%和1.6%,评估指标明显优化。
With the large-scale integration of photovoltaic power generation into the power grid,the resulting uncertainty of the available power supply at the source end has become increasingly prominent.Accurate prediction of photovoltaic output plays a crucial supporting role in optimizing the allocation of power grid resources and improving photovoltaic consumption capacity.This article studies the key factors that affect photovoltaic output and constructs a photovoltaic output prediction model using historical meteorological characteristic data as input.In practice,the model is optimized from the algorithm level to address the subjective and objective issues that exist.①In response to the problem of small sample size and frequent changes in meteorological characteristics in the historical operation of photovoltaic power plants,leading to many sparse feature samples,the ADASYN adaptive adoption algorithm was introduced to rebalance the dataset;②A photovoltaic output model based on meteorological features was constructed using the XGBoost algorithm and compared with traditional BP neural networks.By comparing the actual historical data prediction results of a photovoltaic power plant,combined with ADASYN oversampling and XGBoost algorithm,the accuracy of the model can be effectively improved.Compared to the BP neural networks,MAE,RMSE,MAPE and R2 of the ADASYN-XGBoost algorithm has improved by 66.7%,68.9%,58.0% and 1.6%,respectively,indicating that the evaluation indicators are significantly optimized.
作者
凌煦
周晓刚
陈文哲
符向前
黄社华
LING Xu;ZHOU Xiao-gang;CHEN Wen-zhe;FU Xiang-qian;HUANG She-hua(Central China Branch of State Grid Corporation of China,Wuhan 430072,Hubei Province,China;School of Power and Mechanical Engineering,Wuhan University,Wuhan 430072,Hubei Province,China;School of Water Resources and Hydropower,Wuhan University,Wuhan 430072,Hubei Province,China)
出处
《中国农村水利水电》
北大核心
2024年第6期266-270,共5页
China Rural Water and Hydropower
基金
国家电网科技项目(52140023000U)。
