Machine learning techniques to estimate mechanised forest cutting productivity

The productivity of wood harvesting operations is one the main viability indicators forestry enterprise, which directly influenced by land, population, and operational planning characteristics. variables that affect machines are particularly difficult to measure have complex relationships, making it challenging predict operations. This study generated a model using machine learning (ML) techniques estimate in Eucalyptus plantations southeastern Brazil. input for modelling were average individual tree volumes, volume stand, cutting age, spacing, operator experience, management regime. database was randomly divided into training (70%) validation (30%) datasets. Boosted, artificial neural network (ANN), adaptive network-based fuzzy inference system (ANFIS) used fit evaluated through statistics graphical analysis residues. configurations selected harvester resulted correlation coefficient values greater than 0.9, root-mean-square error (RMSE) percentages less 12.41, indicating strong high accuracy between estimates observed values. boosted technique yielded best results, with coefficients 0.98 0.97, RMSE 6.15 6.65 validation, respectively. worst performance estimating obtained ANFIS technique. ML efficient mechanised forest model.