Scholarlink

<p>This paper proposes a solution to the long-term demand forecasting challenges caused by data scarcity stemming from the growth of economy and electric vehicles. The difficulties caused by the absence of long-term economic and electric-vehicle data were addressed through a novel forecasting model that combines decomposition-linear and exponential-smoothing methods to generate future-input datasets. This forecasting model was validated through a comparison of <i>the original</i> 2022 historical data with <i>the model-generated data</i> for 2022. Next, the <i>model-generated data</i> for 2030 demand was used to increase the long-term forecasting accuracy by effectively capturing complex periodic seasonality and non-decreasing trends. An operating-reserve demand curve was generated based on the long-term forecast data, and the practical use of long-term forecasted demand in reserve planning for future power systems was demonstrated. Our novel model offers a reliable method for addressing data deficiencies in long-term demand forecasting and enables more informed decisions.</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

Long-Term Demand Forecasting: Novel Decomposition-Linear Exponential-Smoothing Algorithm with Long Short-Term Memory

  • 期刊论文

  • 发表日期：

    2025年3月31日

查看全文
  • Tae-Ung Go,
  • Young-Jun Go,
  • Jung-Hyeop Im,
  • Duehee Lee

摘要

This paper proposes a solution to the long-term demand forecasting challenges caused by data scarcity stemming from the growth of economy and electric vehicles. The difficulties caused by the absence of long-term economic and electric-vehicle data were addressed through a novel forecasting model that combines decomposition-linear and exponential-smoothing methods to generate future-input datasets. This forecasting model was validated through a comparison of the original 2022 historical data with the model-generated data for 2022. Next, the model-generated data for 2030 demand was used to increase the long-term forecasting accuracy by effectively capturing complex periodic seasonality and non-decreasing trends. An operating-reserve demand curve was generated based on the long-term forecast data, and the practical use of long-term forecasted demand in reserve planning for future power systems was demonstrated. Our novel model offers a reliable method for addressing data deficiencies in long-term demand forecasting and enables more informed decisions.