一种新型光伏能源电力中长期功率波动预测模型构建  

Construction of a Novel Medium and Long-Term Power Fluctuation Prediction Model for Photovoltaic Power

作　　者：武光华 李宏胜 李鵾 柳长发 WU Guanghua;LI Hongsheng;LI Kun;LIU Changfa(Marketing Service Center of State Grid Hebei Electric Power Co.,Ltd.,Shijiazhuang 050000,Hebei,China)

机构地区：[1]国网河北省电力有限公司营销服务中心,河北石家庄050000

出　　处：《电网与清洁能源》2025年第1期130-136,共7页Power System and Clean Energy

基　　金：国网河北省电力有限公司科技项目(kj2020-085);国家电网公司总部科技项目(5108-202218280A-2-389-XG)。

摘　　要：光伏发电因受太阳辐射周期、地理环境及各种气象因素变化的影响,而使中长期功率波动有较强的不确定性。构建一种新型光伏能源电力中长期功率波动预测模型。基于光伏电池板辐照强度数据的归一化处理,构建光伏发电功率序列波动基础模型;根据波动不确定性,引入模糊径向基函数网络(radial basis function network,RBF)神经网络,利用模糊属性评估波动性,将模型分为5个层级,完成光伏能源电力中长期功率波动的预测。实验结果表明:该方法预测的均方根误差最小值为0.12 kW、平均绝对偏差最小值为0.11 kW、平均绝对百分比误差最小值为1.5%;中长期功率波动预测范围为-9~6 kW,与实际情况完全相符。证明了所构建模型的应用精度更高,性能更理想。Due to the influence of solar radiation cycle,geographical environments and various meteorological factors on photovoltaic power generation,medium and long-term power fluctuations have strong uncertainties.To tackle this issue,a novel photovoltaic energy power medium and long-term power fluctuation prediction model is constructed in this paper.Based on the normalization of the irradiation intensity data of photovoltaic panels,the basic model of photovoltaic power generation power sequence fluctuation is constructed.According to the fluctuation uncertainty,the fuzzy RBF neural network is introduced,and the fluctuation is evaluated by using fuzzy attributes.The model is divided into five levels to complete the prediction of medium and long-term power fluctuations of photovoltaic energy power.The experimental results show that the minimum value of root mean square error predicted by the proposed method is 0.12 kW,the minimum value of average absolute deviation is 0.11 kW and the minimum value of average absolute percentage error is 1.5%.The medium and long-term power fluctuation prediction range is-9~6 kW,which is completely consistent with the actual situation.The results also prove that the application precision of the proposed model is higher and the performance is more ideal.

关 键 词：新型光伏能源 电力中长期功率 波动不确定 预测模型构建 模糊RBF神经网络 

分 类 号：TM715[电气工程—电力系统及自动化]