摘要
针对短期负荷预测精度不够,达到提高系统充裕性评估准确度的目的,本文以呼和浩特市(呼市)地区电网为例,研究了基于大数据的反向传播神经网络(BP神经网络)负荷的短期预测方法。首先,研究了呼市地区电力负荷特性,发现呼市电力负荷变化与温度、节假日等因素相关性;然后,考虑到多重因素对呼市地区负荷变化的影响,以BP神经网络方法为基础,利用大数据主元处理法建立短期电网负荷的预测模型;最后,以呼市地区历史负荷数据为例,通过与传统BP神经网络预测相对比,结果表明基于大数据的BP神经网络短期负荷预测方法的学习时间短、收敛性好、精度高,降低了负荷预测误差,弥补了传统BP神经网络算法的缺点,满足呼和浩特供电局对负荷预测精度要求,提高了系统充裕性评估准确度。
In view of the insufficient accuracy of short-term load forecasting,the purpose of improving the accuracy of projected assessment of system adequacy is achieved,this paper takes the power network of Hohhot as an example,the short-term load forecasting method of back propagation neural network( BP neural network) based on big data is studied. Firstly,the power load characteristics of Hohhot are studied,and the correlation between the change of power load and temperature,holidays and other factors is found. Then,considering the influence of multiple factors on the load change in Hohhot. Based on the BP neural network method,the short-term load forecasting model is established by using the big data principal component processing method. Finally,the historical load data of Hohhot is taken as an example to compare with the forecast of BP network. The results show that the BP neural network based on big data has the advantages of short learning time,good convergence and high accuracy,which can reduce the load forecasting error,make up the shortcomings of the traditional BP neural network algorithm,meet the requirements of Hohhot power supply company for load forecasting accuracy. The accuracy of projected assessment of system adequacy is improved.
作者
姜海洋
周芮冰
王烁罡
周定均
刘昌新
云卿
JIANG Haiyang;ZHOU Ruibing;WANG Shuogang;ZHOU Dingjun;LIU Changxin;YUN Qing(Hohhot Power Supply Breu,Hoho 000 Inner Mongolia,China)
出处
《电力大数据》
2020年第11期47-54,共8页
Power Systems and Big Data
关键词
神经网络法
负荷特性
大数据
主元分析法
短期负荷预测
neural network
power load characteristics
big data
principal component analysis
short-term load forecasting