地基云图结合径向基函数人工神经网络的光伏功率超短期预测模型.docx

DOI：10.13334/j.0258-8013.pcsee.2015.03.007文章编号：0258-8013 (2015) 03-0561-07中图分类号：TM 615 DOI：10.13334/j.0258-8013.pcsee.2015.03.007 文章编号：0258-8013 (2015) 03-0561-07 中图分类号：TM 615 地基云图结合径向基函数人工神经网络的 光伏功率超短期预测模型 陈志宝，丁杰，周海，程序，朱想 (中国电力科学研究院，江苏省 南京市 210003) A Model of Very Short-term Photovoltaic Power Forecasting Based on Ground-based Cloud Images and RBF Neural Network CHEN Zhibao, DING Jie, ZHOU Hai, CHENG Xu, ZHU Xiang (China Electric Power Research Institute, Nanjing 210003, Jiangsu Province, China) ABSTRACT: Due to many local random factors’ effect, the very short-term photovoltaic power forecasting is facing great challenges. Cloud is one of the main factors that makes the surface irradiance fluctuate randomly, thereby causing random changes in output of photovoltaic power, so the cloud need to be quantified and taken into account in the modeling of photovoltaic power forecasting. Firstly, based on all-sky cloud images, the image features related to ground irradiance were extracted using digital image processing techniques. And then the radical basis function (RBF) neural network forecasting model was established, in which the input factors consist of extraterrestrial irradiation, air mass and cloud image features such as image brightness and cloud amount, and the surface irradiance was as output factor. Finally, the very short-term photovoltaic power forecasting was achieved by conversion model of irradiance and power. The experimental results show that, the performance of photovoltaic power forecasting model taking into account the cloud image information was better than the model without any image information. So an important approach was proposed for very short-term photovoltaic power precisely forecast. 征；然后，将大气层外辐射、大气质量、图像亮度和云量作 为输入因子，将地表辐射作为输出，建立径向基函数神经网 络预测模型；最后，根据光电转换模型最终实现光伏功率超 短期预测。实验结果表明：计及地基云图信息的光伏功率超 短期预测模型，效果明显优于无图像信息的模型，为光伏电 站超短期功率精确预测提供了重要的方法。 关键词：地基云图；人工神经网络；光伏功率预测；超短期 0 引言 光伏电站发电功率因受诸多气象因素，环境因 素影响而呈现出固有的随机性和波动性，大规模光 伏发电接入电网会给系统的功率平衡和电网安全 运行带来挑战。因