Otimização de parâmetros de usinagem com machine learning: desenvolvimento de um programa para ajuste de parâmetros conforme a demanda e requisitos de acabamento

Abstract

This work aimed to apply linear regression techniques, using the Orange Data Mining software, to analyze the influence of cutting parameters — depth of cut, feed rate, and cutting speed — on the performance variables of the machining process of cast iron, namely energy consumption, average surface roughness, and material removal rate. Regression models were developed for different insert geometries, whose results presented high determination coefficients (R²), especially for energy consumption (R² = 0.971 for the square insert and R² = 0.962 for the round insert). The analysis of the normalized coefficients allowed identifying feed rate and cutting speed as the most determining factors, varying according to the response variable considered. As a practical application of the results, a web tool (spadini.vercel.app) was developed, providing the models in an intuitive and accessible way, allowing operators and production planners to select cutting parameters according to the variable of interest, supporting decision-making and bridging academic research with industrial practice.