Scholarlink

<p>Power system balancing becomes ever-challenging due to the increasing penetration of uncertain renewable energy sources. Therefore, balancing markets are implemented in various power markets and such markets play a crucial role in ensuring balanced system operation. The optimal operation of balancing market necessitates accurate imbalance volume forecasts. However, imbalance volume forecasting got less research attention. Also, most of the existing imbalance volume forecasting models do not use specific input selection or data preprocessing techniques. These techniques can help to improve forecasting accuracy. Further, balancing market operations necessitates short-term forecasts. Therefore, this paper proposes a novel dynamic regression-based Trigonometric Seasonal, Yeo-Johnson Transformation, ARMA residuals, Trend, and Seasonality (TYJATS) forecasting model to obtain accurate short-term imbalance volume forecasts. The proposed forecasting model dynamically updates data at each time step, effectively handling seasonality and nonlinearity. It uses Grey Correlation Analysis (GCA) for input data selection and the <i>tsrobprep</i> package for data preprocessing. Further, a Newton–Raphson-based optimizer (NRBO) is proposed to minimize the balancing costs. The proposed model is tested on the data collected from the Belgian market. The forecasting analysis demonstrates that the proposed model improves accuracy by 59.85%, 42.61%, and 39.56% compared to machine learning (long short-term memory), regression (auto-regressive distributed lag), and Transmission System Operator (TSO)-based forecasting, respectively. Furthermore, integrating the proposed forecasting model into balancing market optimization results in the least balancing cost, achieving savings of 44.51% compared to TSO-based forecasting.</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

A novel dynamic imbalance volume forecasting model for balancing market optimization

  • 期刊论文

  • 发表日期：

    2025年8月27日

查看全文
  • Chandransh Singh,
  • Sreenu Sreekumar,
  • Tanmoy Malakar

摘要

Power system balancing becomes ever-challenging due to the increasing penetration of uncertain renewable energy sources. Therefore, balancing markets are implemented in various power markets and such markets play a crucial role in ensuring balanced system operation. The optimal operation of balancing market necessitates accurate imbalance volume forecasts. However, imbalance volume forecasting got less research attention. Also, most of the existing imbalance volume forecasting models do not use specific input selection or data preprocessing techniques. These techniques can help to improve forecasting accuracy. Further, balancing market operations necessitates short-term forecasts. Therefore, this paper proposes a novel dynamic regression-based Trigonometric Seasonal, Yeo-Johnson Transformation, ARMA residuals, Trend, and Seasonality (TYJATS) forecasting model to obtain accurate short-term imbalance volume forecasts. The proposed forecasting model dynamically updates data at each time step, effectively handling seasonality and nonlinearity. It uses Grey Correlation Analysis (GCA) for input data selection and the tsrobprep package for data preprocessing. Further, a Newton–Raphson-based optimizer (NRBO) is proposed to minimize the balancing costs. The proposed model is tested on the data collected from the Belgian market. The forecasting analysis demonstrates that the proposed model improves accuracy by 59.85%, 42.61%, and 39.56% compared to machine learning (long short-term memory), regression (auto-regressive distributed lag), and Transmission System Operator (TSO)-based forecasting, respectively. Furthermore, integrating the proposed forecasting model into balancing market optimization results in the least balancing cost, achieving savings of 44.51% compared to TSO-based forecasting.