Predictive model for thick steel laser cutting quality using artificial neural networks

Laser assisted oxygen (LASOX) cutting is emerging as a promising technology for cutting thick mild steel in the industry. The importance of lasox cutting has motivated the researchers to search how effectively good cutting can be obtained for higher thickness of low carbon steel. In this paper, optimum lasox cutting parameters of carbon steel have been determined to achieve minimum Heat Affected Zone (HAZ), minimum kerf width and minimum surface roughness. Cutting experiments have been performed based on statistical experimental design technique. The effects of cutting speed, gas pressure, stand off distance and laser power on the quality of cutting low carbon steel specimens by lasox process have been studied. It is found from the study, that the gas pressure and cutting speed had pronounced effect on cut quality. Low gas pressure produces lower HAZ width, lower kerf width and good surface finish whereas increase in cutting speed results in higher HAZ width, lower kerf width and good surface finish. A predictive model for cutting quality is created using a feed forward artificial neural network exploiting experimental data.