Enhancing biogas production from municipal wastewater sludge and grease trap waste: Explainable machine learning models for prediction and parameter identification

YALÇINKAYA S., Yucel O.

Fuel, cilt.391, 2025 (SCI-Expanded) identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 391
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1016/j.fuel.2025.134787
  • Dergi Adı: Fuel
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, PASCAL, Aerospace Database, Biotechnology Research Abstracts, Chemical Abstracts Core, Chimica, Communication Abstracts, Compendex, INSPEC, Metadex, Pollution Abstracts, Civil Engineering Abstracts
  • Anahtar Kelimeler: Anaerobic digestion, Biogas, Explainable machine learning, Feature importance, Feature selection, Machine learning
  • Marmara Üniversitesi Adresli: Evet

Özet

This research investigates the use of explainable machine learning (ML) models to predict biogas production and identify critical parameters in the anaerobic co-digestion (AcoD) of municipal wastewater sludge (MSS) and grease trap waste (GTW). While traditional ML models offer high accuracy, they often lack transparency, making it difficult to understand the underlying relationships between input variables and outputs. This study addresses this gap by employing various explainable ML approaches. GTW ratio, TS reactor, pH feed, and alkalinity reactor were identified as the most influential parameters for biogas production using SHAP, a superior feature importance method. Sequential backward, F test and Increase in node purity methods selected a larger set of variables with reduced interpretability. Ensemble Least Squares Boosting (LSBoost) with all variables achieved the highest R2 value (0.9555), but Gaussian Process (GP) with SHAP-selected variables showed better performance (R2 = 0.9577) with a simpler model. SHAP effectively identified key parameters, leading to a transparent and generalizable model. This approach balances prediction accuracy with model explainability, aiding in process optimization and control. Overall, this study demonstrates the effectiveness of explainable ML in AcoD modeling. By combining accurate prediction with interpretability, these models offer valuable insights into the complex dynamics of the process and pave the way for improved biogas production and process stability.