Study of the influence of the physical properties of particleboard type P2 on the internal bond of panels using artificial neural networks

Abstract Particleboard panels are normally manufactured in three layers of different sized particles of wood. One of the most important properties of particleboard is internal bond. This study determined the thickness and the physical properties of swelling, water absorption and density of 300 type P2 particleboards, as well as tensile strength perpendicular to the plane of the board, to examine the influence of these physical properties on internal bond of panels. To study the influence on internal bond, a multilayer perceptron artificial neural network was designed using the hyperbolic tangent sigmoid as the transfer function. The artificial neural network designed is capable of explaining at least 82% of the variability of the samples, with no significant differences between the experimental values and those obtained through the network for a significance level of 95%. The neural network proposed is suitable for studying the influence of the physical properties on internal bond, revealing a decrease in internal bond as panel thickness increases. A slight increase in internal bond was observed as swelling and absorption increase to values close to the mean, followed by a decrease. In relation to density, internal bond increases to values of about 700 kg/m 3 , then decreases.