Online SARIMA applied for short-term electricity load forecasting

被引：13

作者：

Anh, Nguyen Thi Ngoc ^{[1
]}

Anh, Nguyen Nhat ^{[2
]}

Thang, Tran Ngoc ^{[1
]}

Solanki, Vijender Kumar ^{[3
]}

Crespo, Ruben Gonzalez ^{[4
]}

Dat, Nguyen Quang ^{[1
]}

机构：

[1] Hanoi Univ Sci & Technol, Hanoi, Vietnam

[2] CMC ATI, 11,Duy Tan St, Hanoi, Vietnam

[3] CMR Inst Technol, Dept Comp Sci & Engn, Hyderabad, TS, India

[4] Univ Int Rioja, Dept Comp Sci & Technol, Logrono, Spain

来源：

APPLIED INTELLIGENCE | 2024年 / 54卷 / 01期

关键词：

Time series; Online SARIMA; Short term forecast; Electricity load forecasting; Online processing; TIME-SERIES; NEURAL-NETWORK; HYBRID ARIMA; MODEL; DEMAND; ALGORITHMS;

D O I：

10.1007/s10489-023-05230-y

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

Short-term Load Forecasting (STLF) plays a crucial role in balancing the supply and demand of load dispatching operations and ensures stability for the power system. With the advancement of real-time smart sensors in power systems, it is of great significance to develop techniques to handle data streams on-the-fly to improve operational efficiency. In this paper, we propose an online variant of Seasonal Autoregressive Integrated Moving Average (SARIMA) to forecast electricity load sequentially. The proposed model is utilized to forecast the hourly electricity load of northern Vietnam and achieves a mean absolute percentage error (MAPE) of 4.57%.

引用

页码：1003 / 1019

页数：17

共 48 条

[1] Kernel-based online machine learning and support vector reduction [J].