基于AdaBoost算法的光伏电站中长期发电预测  

Medium and long-term generation capacity prediction of photovoltaic plantsbased on AdaBoost algorithm

作　　者：王晓东[1] 盛庆博 孙立群 刘绍鹏 王新燕[1] 刘杰[1] WANG Xiaodong;SHENG Qingbo;SUN Liqun;LIU Shaopeng;WANG Xinyan;LIU Jie(Technology Testing Center,Shengli Oilfield CO.,LTD.,Sinopec,Shandong Dongying 257000,China;College of Control and Science Engineering,China University of Petroleum(East China),Shandong Qingdao 266580,China)

机构地区：[1]中国石化胜利油田有限公司技术检测中心,山东东营257000 [2]中国石油大学(华东)控制科学与工程学院,山东青岛266580

出　　处：《工业仪表与自动化装置》2023年第2期65-69,共5页Industrial Instrumentation & Automation

摘　　要：该文提出了一种基于AdaBoost算法的拟建光伏电站发电量预测方法。根据现有光伏电站的历史气象数据与发电量数据,在利用AdaBoost集成学习算法对传统SVM优化的基础上,对气象因素的天气类型进行分类与识别,进而得到4种天气状态下气象因素与发电量之间的对应关系;利用拟建电站所在地的历史气象数据,根据天气类型自动选择对应的LSTM模型,对拟建光伏电站的发电量进行预测。实验结果表明,与采用单一LSTM模型相比,该文方法预测精度有明显的提高,具有一定的推广价值。This paper proposes a method for forecasting the generation of photovoltaic power plants to be built based on AdaBoost algorithm.According to the historical meteorological data and power generation data of existing photovoltaic power stations,and based on the optimization of traditional SVM using AdaBoost integrated learning algorithm,the method classifies and identifies the weather types of meteorological factors,and then obtains the corresponding relationship between meteorological factors and power generation under four weather conditions.Using the historical meteorological data of the place where the power station to be built is located,the corresponding LSTM model is automatically selected according to the weather type to predict the power generation of the photovoltaic power station to be built.The experimental results show that the prediction accuracy of this method is significantly improved compared with that of single LSTM model,and it has certain popularization value.

关 键 词：发电量预测 光伏电站 ADABOOST算法 

分 类 号：TP273[自动化与计算机技术—检测技术与自动化装置]