Scholarlink

<p>Effective energy management in technologically advanced cities is a paramount challenge with the escalating complexity of urban energy networks and the dynamic uncertainty of consumption profiles. In this paper, a hybrid deep learning framework integrating LeNet and GRU (Gated Recurrent Unit), equipped with Artificial Jellyfish Optimization (AJFO) algorithm, is presented to achieve effective short-term energy demand forecasting and adaptive load control in smart cities. The LeNet block learns spatial correlations between energy sensor networks, and the GRU learns temporal dependencies in consumption patterns. AJFO is used for optimal adjustment of the hyperparameters of the hybrid model to provide greater convergence and performance. Comparative results with existing studies show that the proposed framework achieves the lowest latency (4.1 s), lowest mean absolute error (MAE 3.3 kWh) and highest energy savings (16.2%). These outcomes confirm the effectiveness of combining hybrid deep learning techniques with nature-inspired optimization in smart energy management.</p>

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

Hybrid deep learning-driven smart energy management framework in high-tech cities using Lenet, GRU and AJFO

  • 期刊论文

  • 发表日期：

    2025年9月4日

查看全文
  • Fateh Bahadur Kunwar,
  • Shiney Chib,
  • Nitin Tripathi,
  • Dibakar Sinha,
  • Uttam Waghmode,
  • Rajendra Mohite,
  • Snehal Hole,
  • Yogesh Hole

摘要

Effective energy management in technologically advanced cities is a paramount challenge with the escalating complexity of urban energy networks and the dynamic uncertainty of consumption profiles. In this paper, a hybrid deep learning framework integrating LeNet and GRU (Gated Recurrent Unit), equipped with Artificial Jellyfish Optimization (AJFO) algorithm, is presented to achieve effective short-term energy demand forecasting and adaptive load control in smart cities. The LeNet block learns spatial correlations between energy sensor networks, and the GRU learns temporal dependencies in consumption patterns. AJFO is used for optimal adjustment of the hyperparameters of the hybrid model to provide greater convergence and performance. Comparative results with existing studies show that the proposed framework achieves the lowest latency (4.1 s), lowest mean absolute error (MAE 3.3 kWh) and highest energy savings (16.2%). These outcomes confirm the effectiveness of combining hybrid deep learning techniques with nature-inspired optimization in smart energy management.