摘要
对湖州城市居民用电量进行预测过程中,历史用电量数据显示出较强的波动性与季节性,导致原始模型预测效果不理想,文章引入并改进Markov修正组合模型,将Markov修正残差划分部分改进为不同算子残差划分,并用来修正新陈代谢GM(1,1)、SARIMA、Holt-Winters、LSTM等原始模型。使用DC-Markov、MC-Markov、SC-Markov修正后的组合模型预测湖州市的未来月份城市居民用电数据。结果表明,文章提出的最优化残差划分Markov修正模型预测精度较原始模型有一定程度提高,DC-Markov-Holt-Winters模型在湖州市城市居民用电数据的预测上具有较高的精度。
In the process of predicting the electricity consumption of urban residents in Huzhou,the historical electricity consumption data shows strong volatility and seasonality,which leads to the unsatisfactory prediction effect of the original model.In this paper,the Markov modified combination model is introduced and improved,and the residual division part of the Markov modified residual division is improved to adaptive residual division.It was used to modify the original models of metabolism GM(1,1),SARIMA,Holt-Winters,LSTM,etc.The combined model modified by DC-Markov,MC-Markov and SC-Markov is used to predict the residential electricity consumption data of Huzhou City in the coming months.The results show that the prediction accuracy of the adaptive residual division Markov modified model proposed in this paper is improved to some extent compared with the original model,and the DC-Markov-Holt-Winters model has higher accuracy in the prediction of urban residential electricity consumption data in Huzhou City.
作者
曾孟佳
温柔
施闰虎
黄旭
唐陈宇
ZENG Meng-jia;WEN Rou;SHI Run-hu;HUANG Xu;TANG Chen-yu
出处
《智能城市》
2024年第2期49-53,共5页
Intelligent City
基金
教育部人文社会科学一般项目(20YJCZH005)
浙江省湖州市工业攻关项目(2018GG29)
湖州学院国家级大学生创新创业训练项目(202313287007)。
关键词
残差划分
Markov修正
季节性分析
用电量预测
residual division
Markov correction
seasonal analysis
electricity consumption forecast
