Dimensions overall score 7.0

GitHub Code Pulse

No public code linked for this paper yet.

• Classical isotherm models fail to generalize across heterogeneous multi-sample datasets for hydrogen sorption prediction, with R2 collapsing to 0.089–0.382.limitation 95%
• The proposed physics-informed neural network (PINN) framework achieves an R2 of 0.9544, RMSE of 0.0484 mmol/g, and MAE of 0.0231 mmol/g on a held-out test set.result 95%
• The PINN framework enforces physical constraints including saturation limits, monotonic pressure response, and Van't Hoff temperature dependence via penalty weighting in the loss function.method 90%
• Physics-informed regularization provides a 10-15% cross-lithology generalization advantage over a well-tuned random forest model.result 90%
• A seven-category physics-informed feature engineering scheme generates 62 thermodynamically meaningful descriptors from raw material characterization data.method 85%
• The model's predictions show 98.6% monotonicity satisfaction and produce zero non-physical negative predictions.result 90%
• Hydrogen uptake in coals shows a strong positive correlation (r=0.784) with fixed carbon content.result 90%
• The Redlich-Peterson isotherm model suffers from parameter identifiability issues, with its three parameters exhibiting strong correlations during optimization.limitation 85%

Competitive landscape

Competitor map is still being generated for this paper. Enable generation or check back soon.

Startup potential card