基于气象数据外推法和显著性分析的光伏自适应功率预测模型  

PHOTOVOLTAIC ADAPTIVE POWER PREDICTION MODEL BASED ON METEOROLOGICAL DATA EXTRAPOLATION AND SIGNIFICANCE ANALYSIS

作　　者：王丽婕[1] 张青山 郝颖 周颖 邱敏 孙冲 Wang Lijie;Zhang Qingshan;Hao Ying;Zhou Ying;Qiu Min;Sun Chong(School of Automation,Beijing Information Science&Technology University,Beijing 100192,China;Tangshan Research Institute,Beijing Institute of Technology,Tangshan 063000,China;Beijing Key Laboratory of Demand Side Multi-Energy Carriers Optimization and Interaction Technique(China Electric Power Research Institute),Beijing 100192,China;State Grid Hebei Marketing Service Center,Shijiazhuang 050081,China)

机构地区：[1]北京信息科技大学自动化学院,北京100192 [2]北京理工大学唐山研究院,唐山063000 [3]需求侧多能互补优化与供需互动技术北京重点实验室(中国电力科学研究院有限公司),北京100192 [4]国网河北省电力有限公司营销服务中心,石家庄050081

出　　处：《太阳能学报》2025年第2期317-325,共9页Acta Energiae Solaris Sinica

基　　金：国家电网有限公司科技项目(5108-202218280A-2-379-XG)。

摘　　要：分布式光伏电站装机容量较小,一般不进行实时功率统计,难以直接建立功率预测模型。针对分布式光伏电站间光伏组件安装型号和安装方式多样性的问题,基于气象数据外推法和显著性分析提出一种自适应功率预测模型。首先,利用显著性分析筛选光伏组件标称参数和气象数据外推法特征参数之间的相关性,确定与特征参数显著相关的标称参数集合;然后,建立最小二乘支持向量机模型,拟合标称参数集合和特征参数之间的自适应函数关系;最后,建立自适应功率预测模型,基于光伏组件安装方式将环境温度和辐照度转换为板面温度和辐照度,基于自适应函数选择适合当前光伏组件型号的特征参数,根据气象数据外推法得到预测功率。利用光伏电站实际数据进行验证,结果显示,自适应功率预测模型能够为不同型号的光伏组件选择合适的特征参数,通用性更强,相比于目前流行的简化功率预测模型,多云天气下的预测精度提升约2.34%。Distributed photovoltaic power stations typically possess a relatively small installed capacity and generally do not conduct real-time power statistics,posing challenges to directly establish power prediction models.This paper proposes an adaptive power prediction model based on meteorological data extrapolation and significance analysis aiming at the diversity of installation models and methods for photovoltaic panels in distributed photovoltaic power stations.Firstly,significance analysis is conducted on the nominal parameters of the photovoltaic panel and the characteristic parameters in the meteorological data extrapolation method to determine the set of nominal parameters.Then,establish a least squares support vector machine model to fit the adaptive function relationship between the nominal parameter set and the feature parameters.Finally,an adaptive power prediction model is established,which converts the ambient temperature and irradiance into the temperature and irradiance of the panel based on the installation mode of photovoltaic panels,selects the characteristic parameters suitable for the current photovoltaic cell model based on the adaptive function,and inputs the meteorological data extrapolation method to obtain the predicted power.The research results indicate that the adaptive power prediction model can adaptively select appropriate feature parameters for different types of photovoltaic panels,with stronger universality.Compared with the current popular simplified power prediction model,the prediction accuracy is improved by approximately 2.34%.

关 键 词：光伏发电 功率预测 外推法 显著性分析 自适应函数 

分 类 号：TM615[电气工程—电力系统及自动化]