Scholarlink

The Growing utilization of renewable energy sources and the rising complexity of power systems demand the creation of advanced frameworks of smart grids. In this paper, a machine learning-based framework is proposed for the optimization of smart grid operations using real world data obtained from the U.S. Energy Information Administration (EIA). The suggested method takes advantage of the utilization of machine learning models including Long Short-Term Memory Networks (LSTM), Random Forest, XGBoost, Isolation Forest, and Autoencoders for improving grid stability, predicting loads of energy consumption, and detecting anomalies in power distribution. A comprehensive simulation has been carried out to quantify the efficacy of the proposed framework. Among all models, Random Forest achieved the lowest Mean Absolute Error (MAE), making it the most effective for demand forecasting. The findings will be presented in the final section that demonstrate the effectiveness of these models in improving the efficiency of smart grids. The study also contributes to the development of smart grid technology by incorporating advanced analytics and intelligent decision-making.The paper concludes with a discussion of future work, particularly in enhancing the cybersecurity resilience of smart grids by detecting and mitigating attacks such as False Data Injection (FDI), Denial of Service (DoS), Relay, and Load-Altering Attacks (LAA).

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

Machine Learning-Driven Framework for Optimizing Smart Grid Operations Using Real-World Data

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Mais Nijim,
  • Viswas Kanumuri,
  • Waseem Al-Aqqad,
  • Hisham Al-bataineh,
  • Ayush Goyal,
  • David Hicks,
  • George Tuscano

摘要

The Growing utilization of renewable energy sources and the rising complexity of power systems demand the creation of advanced frameworks of smart grids. In this paper, a machine learning-based framework is proposed for the optimization of smart grid operations using real world data obtained from the U.S. Energy Information Administration (EIA). The suggested method takes advantage of the utilization of machine learning models including Long Short-Term Memory Networks (LSTM), Random Forest, XGBoost, Isolation Forest, and Autoencoders for improving grid stability, predicting loads of energy consumption, and detecting anomalies in power distribution. A comprehensive simulation has been carried out to quantify the efficacy of the proposed framework. Among all models, Random Forest achieved the lowest Mean Absolute Error (MAE), making it the most effective for demand forecasting. The findings will be presented in the final section that demonstrate the effectiveness of these models in improving the efficiency of smart grids. The study also contributes to the development of smart grid technology by incorporating advanced analytics and intelligent decision-making.The paper concludes with a discussion of future work, particularly in enhancing the cybersecurity resilience of smart grids by detecting and mitigating attacks such as False Data Injection (FDI), Denial of Service (DoS), Relay, and Load-Altering Attacks (LAA).