ABSTRACT
Wood is a widely used material in various industries, and its mechanical properties are crucial to determine the final product's performance. Modulus of elasticity (MOE) is one of the most important mechanical properties of wood, which measures its stiffness and ability to resist deformation under applied loads. The objective of this research was to create robust models for predicting the moduli of elasticity (MOE) of wood species in three principal directions under varying moisture levels, using the ultrasonic pulse velocity (UPV) technique. The study employed three modeling techniques, namely multivariable linear regression (MLR), artificial neural network (ANN), and support vector regression (SVR). The research involved developing 72 models using different input variables and methods, and then training and testing them with experimental data. Our findings revealed that the SVR model utilizing the radial basis kernel function (RBF) and Levenberg-Marquardt backpropagation ANN exhibited the lowest mean squared error (MSE) and the highest correlation coefficient, with R2 values of 0.979 and 0.989, respectively. These results suggest that the developed models and the UPV technique could significantly enhance the precision of MOE predictions for wood species, presenting a valuable tool for industries that utilize wood in their products.
Data availability
Some or all data, models, or codes that support the findings of this study are available from the corresponding author upon reasonable request.
Disclosure statement
No potential conflict of interest was reported by the author(s).
Notes
1 Near infrared spectrometry based on partial least squares.
2 Levenberg–Marquardt back propagation algorithm was chosen as the training algorithm.
3 The gradient descent with a momentum back propagation algorithm.
4 ANN model was trained by resilient backpropagation.