基于改进灰色模型的光伏发电预测输入数据计算方法  

Calculation Method for Input Data of Photovoltaic Power Generation Predicion Based on Improved Grey Model

作　　者：文贤馗 何明君 周科 蔡永翔 杨垒臣 方学达 WEN Xiankui;HE Mingjun;ZHOU Ke;CAI Yongxiang;YANG Lichen;FANG Xueda(Electric Power Research Institute of Guizhou Power Grid Co.,Ltd.,Guiyang 550002,Guizhou,China;Liupanshui Power Supply Bureau of Guizhou Power Grid Co.,Ltd.,Liupanshui 553000,Guizhou,China)

机构地区：[1]贵州电网有限责任公司电力科学研究院,贵州贵阳550002 [2]贵州电网有限责任公司六盘水供电局,贵州六盘水553000

出　　处：《电力大数据》2024年第7期15-21,共7页Power Systems and Big Data

基　　金：贵州省科技创新人才团队(黔科合平台人才-CXTD[2022]008);南方电网有限责任公司科技项目(GZKJXM20222258)。

摘　　要：人工神经网络是光功率预测的主要模型之一,其输入数据的准确性是影响光功率预测精度的主要因素。该文使用实际测量的、准确的天气历史数据,采用灰色模型GM(1,1)来预测当前的天气数据,并选择多个长度的历史数据序列来进行预测,用相对误差平均值来评估对历史数据的拟合效果,然后选择对历史数据拟合效果最好的序列预测的天气数据,将其与天气预报的天气数据进行加权平均来得到人工神经网络的输入数据,而相应的权重根据灰色模型对历史数据的拟合效果来动态调整。最后,对现有光伏电站数据的仿真验证了该文算法的有效性。Artificial neural networks are one of the main models for predicting optical power,and the accuracy of it’s input data is the main factor affecting the accuracy of optical power prediction.This article uses accurate weather data from historical measurements and the grey model GM(1,1)to predict the current weather data.In this process,multiple lengths of historical data sequences are selected for prediction,and the average relative error is used to evaluate the fitting effect of the historical data.Then,the weather data predicted by the sequence with the best fitting effect on historical data is selected and weighted with the weather data from the weather forecast to obtain the input data of the artificial neural network,and the corresponding weights are dynamically adjusted according to the fitting effect of the grey model on historical data.Finally,the simulation of the existing photovoltaic power station data verifies the effectiveness of the proposed algorithm.

关 键 词：光功率预测 神经网络 灰色模型 输入数据 相对误差平均值 

分 类 号：TM615[电气工程—电力系统及自动化]