Artificial intelligence to predict kerf width during CO2 laser cutting of mild steel

Abstract

Laser cutting is an established, thermal-based manufacturing process that can cut thick metal sheets of complex profiles. In the laser cut process, width of kerf is of greater importance and depends on the selection of appropriate parameters. An artificial neural network (ANN) was used in this research to estimate the width of the kerf during CO2 laser cut of mild steel, taking into account three process parameters: laser beam power (P), speed of cut (S) and pressure of gas (p). A multilayer FFNN (feed-forward neural network) was used to build the artificial neural network predictive model of kerf. The artificial neural network model was trained using 14 of the 17 experimental data points, while the other three were utilized for testing. In both training and testing, the average percentage error was 1.72 percent and 1.05 percent, respectively. It was found that both models and the target results had very low error rates.