Scholarlink

When making decisions and maximizing energy production and distribution to satisfy consumer demands—especially in light of shifting demand—accurate forecasting of energy use is crucial. The extensive use of smart meters has completely changed the way time-series data on energy use is gathered, offering fine-grained insights into usage trends. Eight time-series forecasting models—ARIMA, SARIMA, DT, KNN, SVM, ANN, CNN, and LSTM—are thoroughly compared in this research in order to determine which model performs best on the dataset from smart meters. Several statistical metrics are used to evaluate the models, and the analysis of the experimental findings showed that the LSTM model performed the best in terms of prediction, with an RMSE equal to 2.106, a MAPE equal to 0.19, and an MAE equal to 1.599.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Benchmarking Time-Series Forecasting Models for Smart Meter Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Imane Moustati,
  • Noreddine Gherabi,
  • Bahaa Eddine Elbaghazaoui,
  • Mohamed Amnai

摘要

When making decisions and maximizing energy production and distribution to satisfy consumer demands—especially in light of shifting demand—accurate forecasting of energy use is crucial. The extensive use of smart meters has completely changed the way time-series data on energy use is gathered, offering fine-grained insights into usage trends. Eight time-series forecasting models—ARIMA, SARIMA, DT, KNN, SVM, ANN, CNN, and LSTM—are thoroughly compared in this research in order to determine which model performs best on the dataset from smart meters. Several statistical metrics are used to evaluate the models, and the analysis of the experimental findings showed that the LSTM model performed the best in terms of prediction, with an RMSE equal to 2.106, a MAPE equal to 0.19, and an MAE equal to 1.599.