基于分位数回归神经网络的重庆市用电负荷预测分析
徐闻李¹，杨炜明^1,2
（1.重庆工商大学数学与统计学院，重庆 400067；2.重庆工商大学社会经济应用统计重庆市重点实验室，重庆 400067）
摘要： 2022 年，中国多地最高气温突破历史极值，多日极端高温天气造成重庆市出现较大用电缺口。因为长期电力
负荷的预测存在较大的不确定性，所以选取分位数回归神经网络模型对重庆市未来几年的用电负荷做出预测分析，与几
种预测方法对比分析后，明确分位数回归神经网络模型预测准确程度更高，并于样本外预测了2023—2027 年重庆市年
用电负荷。最后基于预测结果提出一系列恢复正常用电负荷增长的措施，为相关电力部门的规划与建设提供建议。
关键词： 重庆市；用电负荷；分位数回归神经网络
中图分类号： TM715；TP18      文献标志码： A      文章编号： 2095-0802-(2023)10-0022-04
Power Load Forecasting of Chongqing Based on Quantile Regression
Neural Network
XU Wenli¹, YANG Weiming^1,2
(1. School of Mathematics and Statistics, Chongqing Technology and Business University, Chongqing 400067, China;
2. Chongqing Key Laboratory of Applied Statistics for Social Economy, Chongqing Technology and Business University,
Chongqing 400067, China)
Abstract: In 2022, the highest temperature in many places in China broke the historical extreme value. Several days of extreme
high temperature caused a large power shortage in Chongqing. Due to the large uncertainty in the prediction of long-term power
load, the quantile regression neural network model was selected to forecast and analyze the power load of Chongqing in the next
few years. After comparing and analyzing with several forecasting methods, it is clear that the quantile regression neural network
model has a higher degree of prediction accuracy. Besides, the annual power load of Chongqing in 2023 - 2027 was predicted out
of the sample. Finally, based on the forecast results, a series of measures to restore normal load growth were put forward to provide
suggestions for the planning and construction of relevant power departments.

Key words: Chongqing; power load; quantile regression neural network