Forwarder maintenance prediction in cut-to-length system: machine learning approach

• Grant number: 22/06622-2
• Support Opportunities: Scholarships in Brazil - Scientific Initiation
• Start date: August 01, 2022
• End date: July 31, 2023
• Field of knowledge: Agronomical Sciences - Forestry Resources and Forestry Engineering - Techniques and Operations in Forestry
• Principal Investigator: Danilo Simões
• Grantee: Thamires da Silva
• Host Institution: Faculdade de Ciências Agronômicas (FCA). Universidade Estadual Paulista (UNESP). Campus de Botucatu. Botucatu , SP, Brazil

Abstract

The application of machine learning techniques allows making predictions based on data, which can generate a better predictive result in estimating the maintenance of self-propelled forest machines, contributing significantly to the rationalization of resources. That said, the objective is to analyze the performance of machine learning algorithms, through regression analysis, as predictors of target variables of forwarder maintenance in Eucalyptus planted forests. The study will be developed from data from the timber harvesting carried out in the mechanized cut-to-length system, in a clear-cut regime, in planted forests located in the Center-West of the state of São Paulo. The database will be structured from a dataset, with different attributes and, consequently, different instances. Subsequently, the construction of predictive models related to maintenance will take place. Then, supervised learning will be applied, with K-fold cross-validation. The data will be divided into training and testing sets, with 90% and 10% of the total instances, respectively. 18 learning algorithms will be executed, and from these, three algorithms with the best prediction performances will be selected. In order to increase the predictive performance, the selected algorithms will have their hyperparameters adjusted by the tune, ensemble, blend, and stack processes. At the end of this process, the predictive performance of each model will be compared, and applied in the training set and in the test set, from the metrics Coefficient of Determination, Root Mean Square Error, Mean Absolute Error, and Mean Absolute Percent Error.(AU)