Scholarlink

<p>Solar and wind energy’s popularity seems to be on the rise and does not show signs of stopping. However, renewable energy resources tend to be unreliable due to unexpected weather. This may lead to an unstable electrical grid, which causes blackouts and additional problems. These issues could be dealt with using multiple energy management systems that are available to help reduce the risk and stabilize the electrical grid. This study aims to examine the capability of these Smart Grids to become smarter and more accurate and whether energy consumption could be forecast in order to use it in small and big renewable energy plants. This study found that to be easily achievable and with great accuracy. In fact, the minimum amount of training data required to achieve good results across all horizon sizes (1, 2, and 7 days) is one year as concluded by the results. However, a small period of two months is enough to forecast small periods from an hour up to 24&#xa0;h. Additionally, the LSTM network was found to be most suitable.</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

Energy consumption predictions using a neural network

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年11月19日

查看全文
  • Bechara Nehme,
  • Emile Hajj,
  • Chadi Nohra

摘要

Solar and wind energy’s popularity seems to be on the rise and does not show signs of stopping. However, renewable energy resources tend to be unreliable due to unexpected weather. This may lead to an unstable electrical grid, which causes blackouts and additional problems. These issues could be dealt with using multiple energy management systems that are available to help reduce the risk and stabilize the electrical grid. This study aims to examine the capability of these Smart Grids to become smarter and more accurate and whether energy consumption could be forecast in order to use it in small and big renewable energy plants. This study found that to be easily achievable and with great accuracy. In fact, the minimum amount of training data required to achieve good results across all horizon sizes (1, 2, and 7 days) is one year as concluded by the results. However, a small period of two months is enough to forecast small periods from an hour up to 24 h. Additionally, the LSTM network was found to be most suitable.