Published: Jan. 1, 2024

Accurate and reliable long-term forecast of solar radiation temperature is essential, particularly for power plants, in order to support planning the sustainable growth green power. Precise prediction improves production decreases costs. Furthermore, alongside a plant, desalination units can be utilized, where output plant directly influences amount fresh water produced. This article evaluates with capacity 25 megawatts that used as source unit. The objective this unit cover freshwater requirements evaluated region. For purpose, proposed Transformer model was utilized conduct forecasting over 10 years period across three data-driven scenarios. results were compared those other models, including Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), CNN-LSTM, Gated Recurrent Unit (GRU), assess applicability model. study focused on scenarios variations input environmental parameters considered It found high correlation between data target parameters, namely Global Horizontal Irradiance (GHI) Temperature at 2 Meters (T2M), enhances accuracy issue examined using multivariate, multi-step, multi-target approach capabilities suggested With following results, demonstrated best performance 10-year time horizon, MAE (Mean Absolute Error) 10.41 GHI (Global Irradiance) MEA 1.12◦C temperature. reasonable conformity accurate predicted values confirms capability forecasting.

Language: Английский