基于历史数据纵向匹配的光伏功率预测方法研究  

Research on Photovoltaic Power Prediction Based on Historical Data Longitudinally Matching

作　　者：贺衬心 王景丹[1] 霍富强[1] 董永超[1] 王留送[1] 

机构地区：[1]许继集团有限公司,河南许昌461000

出　　处：《高压电器》2017年第5期82-87,95,共7页High Voltage Apparatus

摘　　要：针对现有光伏功率预测系统存在预测误差大的问题,文中结合实际工程经验,提出一种基于历史数据纵向匹配的功率预测方法。通过对电站历史数据进行预处理,排除失真数据,以时间和气象预测数据为条件对历史数据进行纵向查询,纵向匹配出预测日的光辐射系数曲线、功率输出综合系数曲线,实现对未来几小时或几天的功率预测。仿真实验与电站实际应用表明:该方法预测误差小,简单实用,有实际推广意义。Aiming at the problems of the existing prediction have large error in the PV power prediction system, combined with practical engineering experience,proposed a power prediction method based on historical data longitudinally matching. To processed the historical data of the power station, for eliminating distortion data. Querying historical data with time and prediction data of weather longitudinally, match out the prediction day＇s curve of the optical radiation coefficient and the power output, to realize the power prediction for the next few hours or days. Simulation experiment and practical application show that this method has small error,simple and practical, and has practical significance.

关 键 词：历史数据 纵向匹配 光辐射系数 功率输出综合系数 功率预测 

分 类 号：TM615[电气工程—电力系统及自动化]