Scholarlink

<p>One of the most important resources for industries worldwide is electricity. As a result, electricity's final price affects the economy. The weather, geography, and operational factors affect the final electricity price. As a result, forecasting and assessing the sensitivity of parameters of the marginal price of power is necessary for managing and planning energy output. In this study, it was paid to predict the final power energy price using the machine learning (ML) techniques eXtreme Gradient Boosting (XGB), CatBoost, and Histogram Gradient Boosting Regression (HGBR). The Archimedes optimization algorithm (ArchOA) and the hunger game search were used to optimize the hyperparameters of each of the three approaches. Subsequently, three basic prediction models and six hybrid prediction models were compared. The results demonstrated that using hybrid models instead of the basic models greatly lowered error value, improved precision, and promptly forecasted. As a result, the performance and accuracy of the XGBoost-ArchOA are the best. The findings of the sensitivity analysis also demonstrate that the price of energy is significantly impacted by fuel prices.</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

An artificial intelligence and sensitivity analysis-based strategy for forecasting marginal price of electricity

  • 期刊论文

  • 发表日期：

    2026年5月15日

查看全文
  • Jie Yuan

摘要

One of the most important resources for industries worldwide is electricity. As a result, electricity's final price affects the economy. The weather, geography, and operational factors affect the final electricity price. As a result, forecasting and assessing the sensitivity of parameters of the marginal price of power is necessary for managing and planning energy output. In this study, it was paid to predict the final power energy price using the machine learning (ML) techniques eXtreme Gradient Boosting (XGB), CatBoost, and Histogram Gradient Boosting Regression (HGBR). The Archimedes optimization algorithm (ArchOA) and the hunger game search were used to optimize the hyperparameters of each of the three approaches. Subsequently, three basic prediction models and six hybrid prediction models were compared. The results demonstrated that using hybrid models instead of the basic models greatly lowered error value, improved precision, and promptly forecasted. As a result, the performance and accuracy of the XGBoost-ArchOA are the best. The findings of the sensitivity analysis also demonstrate that the price of energy is significantly impacted by fuel prices.