光伏发电系统功率预测方法研究综述  

Review of Power Forecasting Methods of Photovoltaic Power Generation System

作　　者：吴硕[1] WU Shuo(Liaoning Equipment Manufacture College of Vocational and Technical,Shenyang,China,Post Code:110161)

机构地区：[1]辽宁装备制造职业技术学院,辽宁沈阳110161

出　　处：《热能动力工程》2021年第8期1-7,共7页Journal of Engineering for Thermal Energy and Power

基　　金：2020年辽宁省教育厅科学研究项目(SYZB202004)。

摘　　要：光伏发电大规模并网给电网的稳定运行带来巨大挑战,提高光伏发电功率预测水平对光伏能源并网安全具有重要意义。光伏发电系统功率输出具有明显的非线性、间接波动性和不确定性特点,须采用数学模型结合多尺度预测方法实现较高预测精度。针对多元线性回归算法、马尔科夫链预测、神经网络算法、支持向量机和组合预测等光伏系统输出功率的直接预测方法,及基于光伏电站精确建模的光伏系统输出功率的间接预测法,从基本思想、技术路线、适用条件和应用效果的角度进行分析与总结。当前功率预测方法主要有基于统计学的某种学习方法和组合预测方法,数据处理加工是关键因素。预测数据可用性的评估标准和预测方法的工程应用是未来研究工作的重点。The large-scale grid-connection of photovoltaic power generation brings great challenges to the stable operation of the grid.Improving the forecast level of photovoltaic power generation is of great significance to the safety of photovoltaic energy grid-connection.The power output of photovoltaic power generation system has obvious characteristics of nonlinearity,indirect volatility and uncertain factors.It is necessary to adopt mathematical models combined with multi-scale forecaasting methods to achieve higher forecasting accuracy.The basics of direct forecasting methods of photovoltaic system output power such as multiple linear regression algorithms,Markov chain forecast,neural network algorithms,support vector machines and combination forscast,and indirect forecasting methods of photovoltaic system output power based on accurate modeling of photovoltaic power plants thoughts,technical routes,applicable conditions and application effects are analyzed and summarized.Current power forecasting methods mainly focus on a certain learning method or combination forecasting method based on statistics,and data processing is the key factor.The evaluation criteria for the availability of forecast data and the engineering application of forecasting methods are the focus of future research work.

关 键 词：光伏系统 功率预测 直接预测法 间接预测法 数据处理 

分 类 号：TM615[电气工程—电力系统及自动化]