.png){kind=icon}
.png){kind=icon}
Return
A scalable and interpretable machine learning framework for optimizing methane yield and process reliability in psychrophilic anaerobic digestion
.png){kind=icon}
OA
AI
A
R
DOI：10.1016/j.ecmx.2026.102054.png)
.png){kind=icon}
.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Developed a reactor-independent ML framework for psychrophilic AD systems. • Benchmarked 11 ML models using a 38-month multi-reactor operational dataset. • ANN achieved highest predictive accuracy (R2 ≈ 0.85) across all models tested. • LORO validation showed ∼ 70% success rate on entirely unseen reactor systems. • HRT, OLR, FAN/TAN ratio and VFAs identified as key drivers of AD stability.
Keywords:
Psychrophilic anaerobic digestion
Methane yield prediction
Leave-One-Reactor-Out validation
Zero-leakage machine learning
SHAP interpretability
Uncertainty quantification
AD
Anaerobic Digestion
SMY
Specific Methane Yield
ANN
Artificial Neural Network
SVM
Support Vector Machine
RF
Random Forest
XGBoost
Extreme Gradient Boosting
DT
Decision Tree
KNN
K-Nearest Neighbors
PSO
Particle Swarm Optimization
GA
Genetic Algorithm
GPR
Gaussian Process Regression
ELM
Extreme Learning Machines
FL
Fuzzy Logic
SHAP
SHapley Additive exPlanations
TAN
Total Ammonia Nitrogen
TKN
Total Kjeldahl Nitrogen
VFA
Volatile Fatty Acids
COD
Chemical Oxygen Demand
TS
Total Solids
VS
Volatile Solids
OLR
Organic Loading Rate
CLR
Carcass Loading Rate
HRT
Hydraulic Retention Time
SBR
Sequencing Batch Reactor
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
E
IF:
10.9
Papers:
902
Citations:
4
Organization
No organization information available
Cited Papers