摘要
精确的电量预测是电网投资和电力平衡的重要依据。为解决以历史用电量数据作为电量预测的唯一依据,在模型迭代过程中的原始信息少,预测模型通用性差、预测精度低等问题,通过对地区月用电量与经济因素进行研究分析,采用随机森林算法对经济因素和月用电量进行针对性建模预测。通过实际算例仿真验证表明,该模型预测MAPE(平均绝对百分误差)为2.34%,预测精度高并且通用性强。
Precise electricity consumption forecasting provides an important basis for the investment of power grid and electric power balance.To solve the problems such as usage of historical consumption data as the only basis for electricity consumption forecasting,little information in the process of iteration model,poor universality and low accuracy of a forecasting model,this paper adopts random forest to model and forecast economic factors and monthly electricity consumption.The simulation and verification of examples show that the model's MAPE(mean absolute percentage error)is 2.34%,which obtains high prediction accuracy and universality.
作者
屠一艳
徐久益
杨晓雷
李自明
姚剑峰
TU Yiyan;XU Jiuyi;YANG Xiaolei;LI Ziming;YAO Jianfeng(State Grid Jiaxing Electric Power Supply Company,Jiaxing Zhejiang 314000,China;State Grid Tongxiang Power Supply Company,Tongxiang Zhejiang 314500,China)
出处
《浙江电力》
2021年第3期91-96,共6页
Zhejiang Electric Power
基金
国家电网有限公司质量管理创新基金项目(JDTX-2020Q)。
关键词
电量预测
经济因素
数据挖掘
算法模型
随机森林
electricity consumption prediction
economic factor
data mining
algorithm model
random forests
