Data-Driven Modelling of Carbon Nanotube Reinforced Composite Plates

Surya Dev Singh published a dataset of 144 plate configurations for functionally graded carbon nanotube-reinforced composite plates on figshare in May 2026. The data was generated by varying loading conditions, CNT volume fractions, width-to-thickness ratios, and CNT distribution patterns to predict static bending behavior. Fifteen machine learning models were trained on this data to estimate central deflection, normal stress, in-plane shear stress, and transverse shear stress.

Use Cases

• Training surrogate models to predict central deflection based on width-to-thickness ratio and loading conditions.
• Analyzing the influence of CNT volume fraction and distribution patterns on stress behavior.
• Comparing the performance of ensemble models like Gradient Boosting Regressor against linear models for nonlinear composite systems.
• Conducting sensitivity analysis to determine optimal model parameters like the number of estimators.

Strengths

• Dataset contains 144 distinct plate configurations.
• Model performance is documented, with Gradient Boosting Regressor achieving R² up to 0.9996.
• The study includes feature importance analysis, identifying width-to-thickness ratio as a primary governing factor.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• The dataset is small at 40.7 KB, indicating a limited scope for training generalizable models.

Provenance

Source

Surya Dev Singh

Collection Method

Generated by varying plate configuration parameters, likely from finite element simulations.

Freshness

Last updated 2026-05-13 05:35:18; freshness should be verified.