Aplicação de técnicas de aprendizado de máquina para determinar a velocidade de produção em máquinas coextrusoras
Due to the need to optimize processes, resources and decisions, industries have invested in information systems errors and losses in their production process. Having a greater reliability accuracy in the information received is vital in decision making, and the use of machine learning has helped the...
| Autor principal: | Bomfim, Marlon Alves |
|---|---|
| Formato: | Trabalho de Conclusão de Curso (Graduação) |
| Idioma: | Português |
| Publicado em: |
Universidade Tecnológica Federal do Paraná
2022
|
| Assuntos: | |
| Acesso em linha: |
http://repositorio.utfpr.edu.br/jspui/handle/1/27536 |
| Tags: |
Adicionar Tag
Sem tags, seja o primeiro a adicionar uma tag!
|
| Resumo: |
Due to the need to optimize processes, resources and decisions, industries have invested in information systems errors and losses in their production process. Having a greater reliability accuracy in the information received is vital in decision making, and the use of machine learning has helped the industries, and for this reason, the implementation of Machine Learning (ML) methods contribute to the gain of results in the data automation and reliability. With this focus, this paper reports the application of ML classification and regression techniques to predict the production speed needed for the manufacture of coextruded materials, based on historical data provided To study this problem, consolidated ML techniques were chosen. The data was preprocessed and we used 12 machine learning techniques in the study of classification and regression models altogether, each one with distinct configurations. The first method aims to classify among historical values which the data base has to define the production speed and setup, and the second method aims to determine the approximate speed using the database as calculation source. The results were satisfactory, with emphasis on the Decision Trees and Random Forest techniques, which obtained an average accuracy rate of 77% and 78% in classification and 79% and 82% in regression, respectively. |
|---|