Machine learning-based prediction of biomass energy potential from agricultural residues in Algeria

Yazar(lar): Mohamedeltayib Omer Salih Eissa, Y. Benal Öztekin, Omsalma Alsadig Adam Gadalla, Geofrey Prudence Baitu, Khaled Adil Dawood Idress

Özet:
The aim of this study was to assess and predict the biomass energy potential derived from agricultural residues in Algeria. Biomass energy generation from agricultural production in Algeria holds significant potential due to the country's vast agricultural resources. Algeria has diverse agricultural activities ranging from cereal cultivation to olive farming, offering various biomass feedstocks for energy production. Given the country's significant agricultural activities, residues such as straw, stalks, and husks from crops like wheat, barley, maize, and potatoes represent a valuable source of bioenergy. Production data for the 2022 growing season were obtained from the FAOSTAT database, and residue quantities were calculated using residue-to-product ratios (RPR) and calorific values. The total amount of agricultural waste was estimated at approximately 15.3 kilotons, corresponding to an energy potential of around 279 terajoules (TJ). To enhance the predictive capacity of this assessment, a machine learning approach was employed using a Random Forest Regressor. The model was trained using crop-specific features such as production volume, RPR, availability, and lower heating value (LHV) to estimate the energy potential of residues. While the model showed a strong ability to capture energy potential trends, evaluation metrics indicated room for optimization (R² = –19693.04, RMSE = 19,438.57 GJ, MAE = 17,149.01 GJ), likely due to limited dataset size. Nevertheless, the integration of ML demonstrates the feasibility of applying data-driven models to estimate biomass energy from agricultural residues and supports future planning and development of renewable energy strategies in Algeria.

DOI: https://doi.org/10.58559/ijes.1796758

Makale Sayfası: DergiPark bağlantısı

Tam Metin (PDF): PDF indir

Kategori: Hakemli