Table 1.
Nomenclature.
Fig 1.
(a) Representative flexural load-displacement curve; (b) schematic of three-point loading test (ASTM D7264).
Fig 2.
(a) Tensile test set-up with a representative load-displacement curve; (b) schematic of standard tensile test method for composite materials (ASTM D3039).
Fig 3.
(a) A representative load-displacement curve obtained from mode II fracture mechanics testing; (b) schematic of end notch flexure specimen.
Fig 4.
Methodology chart illustrating steps of data preparation and model training.
Table 2.
Mean and standard deviation values of glass transition temperature and mechanical properties nine different types of CFRP composites.
Table 3.
Summary of the RMSE and R2 from ridge regression, support vector regression, and random forest based on LOOCV without including the Type as an independent variable. Note: only independent variables were normalized. Asterisks correspond to the highest obtained values with respect to the three models.
Fig 5.
Prediction of flexural strength using ridge regression.
(a) Predicted versus actual values; (b) Feature premutation.
Fig 6.
Prediction of flexural modulus using ridge regression.
(a) Predicted versus actual values; (b) Feature premutation.
Fig 7.
Prediction of mode II energy-release rate using the random forest model without including the Type variable.
(a) Predicted versus actual values; (b) Feature premutation.
Table 4.
Summary of the RMSE and R2 from ridge regression, SVM, and RF based on LOOCV with including the Type as independent variable. Note: only independent variables were normalized. Arrows are used to indicated changes in the values compared with results obtained in Table 3.
Fig 8.
Prediction of mode II Energy Release Rate using ridge regression with including the type.
(a) Predicted versus actual values; (b) Feature premutation.