Artificial intelligence to growth stresses predicting in Eucalyptus clones using dendrometric variables and wood density

Abstract

Eucalyptus planted forests contribute to maximizing lumber production but problems such as longitudinal growth strain can negatively influence the quality of the products. Knowing dendrometric variables and wood properties can help in the prediction of longitudinal growth strain, mainly with the help of artificial intelligence. Thus, the aim of this research was to evaluate the use of artificial neural networks to predict longitudinal growth strain in Eucalyptus trees based on dendrometric variables, spacing between trees and wood density. The longitudinal growth strain was measured in trees of four Eucalyptus clones planted in three spacings. The diameter and height of each tree were measured. The basic wood density was determined. Artificial neural networks were used to estimate longitudinal growth strain as a function of dendrometric variables, tree spacing and wood density. The results showed that the artificial neural networks presented good results for training and validation, with most of them resulting in high correlation coefficient values. The trained artificial neural networks showed a correlation coefficient above 0,56. Artificial neural networks showed that the variables clone and basic wood density were the ones that most contributed to the prediction of longitudinal growth strain. On the other hand, the spacing between trees, the height of the tree and the diameter at breast height were not relevant to predict growth stresses.