Hybrid feature-based neural network regression method for load profiles forecasting

Energy Informatics, Год журнала: 2025, Номер 8(1)

Опубликована: Фев. 10, 2025

This study addresses the critical need for improved demand forecasting models that can accurately predict energy consumption, particularly in context of varying geographical and climatic conditions. The work introduces a novel model integrates clustering techniques feature engineering into neural network regression, with specific focus on incorporating correlations air temperature. Evaluation model's efficacy utilized benchmark dataset from Tetouan, Morocco, where existing methods yielded RMSE values ranging 6429 to 10,220 [MWh]. In contrast, proposed approach achieved significantly lower 5168, indicating its superiority. Subsequent application forecast Astana, Kazakhstan, as case study, showcased further. Comparative analysis against baseline method revealed notable improvement, exhibiting MAPE 5.19% compared baseline's 17.36%. These findings highlight potential enhance accuracy, across diverse contexts, by leveraging climate-related inputs, methodology also demonstrates broader applications, such flood forecasting, agricultural yield prediction, or water resource management.

Язык: Английский

Methods and models of decision-making regarding energy saving in communal facilities based on Delphi method

Опубликована: Май 15, 2024

Язык: Английский