Otimização do controle da transmissão de um trator agrícola híbrido por meio de técnicas de Machine Learning
Saved in:
| Main Author: | |
|---|---|
| Publication Date: | 2025 |
| Format: | Master thesis |
| Language: | por |
| Source: | Repositório Institucional da UPF |
| Download full: | https://repositorio.upf.br/handle/123456789/10014 |
Summary: | The agricultural tractor is one of the main pieces of machinery that a farmer possesses on their farm, being responsible for the majority of agricultural operations. Its technological advancements have brought versatility and performance, revolutionizing the way farmers work and becoming essential components of modern agriculture. This progress, combined with precision agriculture, provides an amount of data that can assist farmers in decisionmaking, especially data extracted from agricultural machinery. In this context, this work proposes the use of performance data from a hybrid agricultural tractor for the development of a transmission control optimization system. This system is based on the use of Machine Learning techniques, utilizing data obtained under different operating conditions and attached implements to suggest new controller gain parameters, aiming for responsiveness and adaptability in the tractor’s operational performance. Data were collected from four scenarios: transport without implement, light harrowing, heavy harrowing, and planting with a planter. A total of 17 Machine Learning models were used to learn from the data in these scenarios and to suggest new parameters. For model evaluation, the metrics used were Mean Absolute Error (MAE) and the coefficient of determination (R²). The Extreme Gradient Boosting (XGBoost) regression model proved to be the most promising, with an MAE of 0.2382 and an R² of 0.9993. The parameters suggested by the model replace the original controller parameters, optimizing the system. |
Be the first to leave a comment!