Scholarlink

AI and IoT-driven smart cities optimize energy consumption, enhance efficiency, and reduce environmental impact through real-time analytics, predictive modeling, and automated distribution to improve grid resilience, integrate renewables, and enable adaptive energy management, ensuring sustainability and economic benefits. In this research, an extensive examination of high-impact scholarly works was explored in the domain of AI-driven energy consumption optimization, which reveals several research lacunae, including the absence of empirical validation, deficiencies in real-world deployment methodologies, and a paucity of comprehensive strategies to surmount challenges such as scalability, interoperability, and seamless AI integration within smart city infrastructures. It underscores the efficacy of machine learning models in forecasting energy consumption, with a pronounced focus on the optimized GBDT++ model, which solves the current research gaps. The Advanced GBDT++ framework undergoes training within the Scikit-learn ecosystem, leveraging Grid Search and Bayesian Optimization for meticulous hyperparameter calibration. The balanced dataset comprises 1100 records, including 387 residential, 351 industrial, and 362 commercial energy consumption entries, with six features (building type, square footage, occupants, appliances used, average temperature, and weekday, measured in megawatts). Comparative evaluations of Decision Tree, SGBoost, AdaBoost, Bagging, and Extra Trees demonstrate that the GBDT++ model emerges as preeminent, attaining an R2 of 0.9947, MAE of 52.92, RMSE of 65.96, and MAPE of 1.36%, signifying exemplary precision and robustness. Future research will focus on real-time implementation, model interpretability enhancement, edge computing, renewable energy forecasting, and reinforcement learning and cybersecurity to create an AI-powered energy optimization framework.

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

A Powerful Machine Learning Approach for Optimizing Energy Consumption in Smart Cities

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Jannatul Ferdousmou,
  • Mahafuj Hassan,
  • Anamika Tiwari,
  • Nur Vanu,
  • Muslima Begom Riipa,
  • Srabani Das,
  • Sumaiya Yeasmin,
  • Md Fakhrul Hasan Bhuiyan

摘要

AI and IoT-driven smart cities optimize energy consumption, enhance efficiency, and reduce environmental impact through real-time analytics, predictive modeling, and automated distribution to improve grid resilience, integrate renewables, and enable adaptive energy management, ensuring sustainability and economic benefits. In this research, an extensive examination of high-impact scholarly works was explored in the domain of AI-driven energy consumption optimization, which reveals several research lacunae, including the absence of empirical validation, deficiencies in real-world deployment methodologies, and a paucity of comprehensive strategies to surmount challenges such as scalability, interoperability, and seamless AI integration within smart city infrastructures. It underscores the efficacy of machine learning models in forecasting energy consumption, with a pronounced focus on the optimized GBDT++ model, which solves the current research gaps. The Advanced GBDT++ framework undergoes training within the Scikit-learn ecosystem, leveraging Grid Search and Bayesian Optimization for meticulous hyperparameter calibration. The balanced dataset comprises 1100 records, including 387 residential, 351 industrial, and 362 commercial energy consumption entries, with six features (building type, square footage, occupants, appliances used, average temperature, and weekday, measured in megawatts). Comparative evaluations of Decision Tree, SGBoost, AdaBoost, Bagging, and Extra Trees demonstrate that the GBDT++ model emerges as preeminent, attaining an R2 of 0.9947, MAE of 52.92, RMSE of 65.96, and MAPE of 1.36%, signifying exemplary precision and robustness. Future research will focus on real-time implementation, model interpretability enhancement, edge computing, renewable energy forecasting, and reinforcement learning and cybersecurity to create an AI-powered energy optimization framework.