光伏发电预测方法研究进展  

A Review of Photovoltaic Power Generation Forecasting Methods

作　　者：舒胜 谢应明[1] 杨文宇 邹俊华 SHU Sheng;XIE Ying-ming;YANG Wen-yu;ZOU Jun-hua(School of Energy and Power Engineering,University of Shanghai for Science and Technology,Shanghai,China,200093)

机构地区：[1]上海理工大学能源与动力工程学院,上海200093

出　　处：《热能动力工程》2020年第11期1-11,共11页Journal of Engineering for Thermal Energy and Power

基　　金：国家自然科学基金(50806050)。

摘　　要：光伏发电大规模接入电网会使电网产生一定波动,对电力系统产生影响,提高光伏发电量预测的准确性是发展光伏发电技术及保证电网稳定性的关键。本文对光伏发电量预测方法进行归纳总结,根据研究原理将其分为直接预测法和间接预测法,并对直接预测法中的混合模型做了具体分类:基于确定神经网络初始权值的混合模型、基于光伏数据预处理的混合模型及其他混合模型。通过比较各种方法的平均绝对百分比误差(MAPE)及仿真时间,对各种方法进行评估。结果表明:人工智能预测法目前应用最广,MAPE在3%~15%之间,其中,深度学习网络模型预测误差最小,但仿真时间较长且模型复杂度较高;混合模型可以有效减小预测误差,总体预测误差小于10%,是未来一个重要的研究领域。Large-scale access to photovoltaic power generation will cause certain fluctuations of power grid and affect the power system.Therefore,improving the accuracy of photovoltaic power generation forecast is the key to developing photovoltaic power generation technology and ensuring grid stability.In this work,the photovoltaic power generation forecasting methods are categorized and summarized.According to the research principle,it is divided into two categories:direct and indirect forecasting methods.And hybrid model is divided into three types:hybrid model based on determining initial weights of neural networks,hybrid model based on the preprocessing of photovoltaic data and other hybrid model.And the various methods were evaluated by comparing the mean absolute percentage error(MAPE)and simulation time.Our review shows that the artificial intelligence forecasting method is the most widely used,and the MAPE is between 3%and 15%.The forecasting error of the deep learning network model is the smallest,but the simulation time is longer and the model complexity is higher.The hybrid model can effectively reduce the forecasting error.Its forecasting error is less than 10%,which is an important research area in the future.

关 键 词：光伏发电 预测方法 人工智能 混合模型 

分 类 号：TM615[电气工程—电力系统及自动化]