基于XGBoost算法的用电电量预测的实践应用
|
摘要
用电电量预测是电力供应单位的重要工作之一,对于分析地区行业的经济发展趋势着有重要作用。由于用电需求受多种因素的共同影响,电量的预测区间不同,对预测结果精度要求各有不同。为了改善预测结果,提出基于XGBoost算法构建电力预测模型,预测工业用户的短期电量及行业的中长期电量。以广东省某地级市市2014年-2017年上半年的实际数据,采取工业用户不同时间序列的数据建模进行训练并与真实值比较验证模型的可靠性,并对2017年下半年的工业用户的月电量进行预测。结果表明:模型能够在预测地区行业的长期电量上有较高的精度;在误差可控的情况下,预测工业用户的短期电量。
The prediction of electricity consumption is one of the important work of the power supply enterprise,and plays an important role in analyzing the economic development trend of the regional industry.As the demand of electricity is influenced by many factors,the forecast interval of electric energy is different,and the accuracy of prediction result is different. In order to improve the prediction results,a power forecasting model based on XGBoost algorithm is proposed to predict the short-term electricity consumption of industrial users and the medium and long-term power consumption of the industry. Based on the actual data of a prefecture level city in Guangdong in the first half of 2014-2017,the data model of industrial users with different time series is trained and compared with the real value to verify the reliability of the model,and the monthly electricity consumption of the industrial users in the second half of 2017 is predicted. The results show that the model can predict the long-term power consumption of the industry with high precision; and the shortterm electricity consumption of industrial users can be predicted under the condition of controllable error.
The prediction of electricity consumption is one of the important work of the power supply enterprise,and plays an important role in analyzing the economic development trend of the regional industry.As the demand of electricity is influenced by many factors,the forecast interval of electric energy is different,and the accuracy of prediction result is different. In order to improve the prediction results,a power forecasting model based on XGBoost algorithm is proposed to predict the short-term electricity consumption of industrial users and the medium and long-term power consumption of the industry. Based on the actual data of a prefecture level city in Guangdong in the first half of 2014-2017,the data model of industrial users with different time series is trained and compared with the real value to verify the reliability of the model,and the monthly electricity consumption of the industrial users in the second half of 2017 is predicted. The results show that the model can predict the long-term power consumption of the industry with high precision; and the shortterm electricity consumption of industrial users can be predicted under the condition of controllable error.
引文
[1]李艳红,雷金辉.电力负荷时间序列预测的应用与研究[J].科学技术与工程,2011,11(4):860-864.
[2]张华,刘素兵,王智.基于小波的ARMA模型对我国电力的预测研究[J].河南科技,2013,11(22):49.
[3]罗序良.基于主成分分析的多元回归算法在电力预测中的应用[J].机电信息,2016(36):20-21.
[4]蒋喆.支持向量机在电力负荷预测中的应用研究[J].计算机仿真,2010,27(8):282-285.
[5]王吉权,王福林,董志贵,等.基于改进BP神经网络的电力负荷预测[J].数学的实践与认识,2017,47(9):276-284.
[6]马立新,李渊.区间电力负荷特征提取及预测方法[J].控制工程,2015,22(4):645-648.
[2]张华,刘素兵,王智.基于小波的ARMA模型对我国电力的预测研究[J].河南科技,2013,11(22):49.
[3]罗序良.基于主成分分析的多元回归算法在电力预测中的应用[J].机电信息,2016(36):20-21.
[4]蒋喆.支持向量机在电力负荷预测中的应用研究[J].计算机仿真,2010,27(8):282-285.
[5]王吉权,王福林,董志贵,等.基于改进BP神经网络的电力负荷预测[J].数学的实践与认识,2017,47(9):276-284.
[6]马立新,李渊.区间电力负荷特征提取及预测方法[J].控制工程,2015,22(4):645-648.