본문 바로가기
  • Home

Development of an AI-Based Energy Management System for Factory Power Saving

  • Journal of Internet of Things and Convergence
  • Abbr : JKIOTS
  • 2024, 10(6), pp.49-55
  • Publisher : The Korea Internet of Things Society
  • Research Area : Engineering > Computer Science > Internet Information Processing
  • Received : November 1, 2024
  • Accepted : December 13, 2024
  • Published : December 31, 2024

Ilyosbek Rakhimjon-Ugli Numonov 1 Bo Peng 1 Yanxia Li 2 Yuldashev Izzatillo Hakimjon Ugli 2 TaeO Lee 1 KIM, TAEKOOK 1

1국립부경대학교
2국립부경대학교 컴퓨터공학과

Accredited

ABSTRACT

In this paper, AI models for predicting peak power usage were developed and comparatively analyzed using data collected from the Jeju Samdasoo factory through a big data collection system based on IoT sensing technology. The LSTM (Long Short-Term Memory) model demonstrated the highest prediction accuracy for univariate time-series data, achieving an R² of 0.98, RMSE of 0.039, and MAE of 0.026. Meanwhile, the XGBoost (eXtreme Gradient Boosting) model effectively handled multivariate data, achieving an R² of 0.93, RMSE of 0.018, and MAE of 0.013. Various data preprocessing methods and feature combinations were experimentally applied to optimize model performance, highlighting the significant impact of preprocessing and variable selection on prediction accuracy. The findings suggest that optimized AI models for peak power prediction can reduce power costs and achieve approximately 10–15% reductions in carbon emissions. This study offers companies pursuing ESG (environmental, social, and governance) management practical and specific strategies for achieving sustainability, while demonstrating the applicability of the predictive model across various industries, including manufacturing, logistics, and smart factories.

Citation status

* References for papers published after 2023 are currently being built.