Modeling the evaluation of methods for determining the basic density of wood in forest species based on data from a neuro-fuzzy inference system

Abstract

The forestry sector is one of the agribusiness sectors that generates the most wealth for the national economy, as it brings benefits to society, from the wood itself for industries, biomass for energy production, and to the environment, reducing pressure on native forests and the reuse of land degraded by agriculture. In view of this, this study was carried out to predict the different basic densities in tree species under the influence of two factors, nine different tree species in relation to three different density methodologies using the Neuro-Fuzzy System. Tree basic density modeling was carried out using effective species parameters and different calculation methodologies adapted to the Neuro-Fuzzy Inference System (ANFIS). In the ANFIS model, 67% and 33% of the total data were considered as training and test data, respectively. The numbers of pertinence functions were selected 9 for species and 3 for methodologies for the input data. ANFIS training was carried out using the hybrid method. The average R2 determination coefficients were 87.32% and 97.42% for the field and ANFIS models, respectively. The model obtained using ANFIS showed a high accuracy of 4.36%. Compared to the field data, the ANFIS model was highly accurate and can be used to estimate the basic density of the trees in this study.