Kerf profile analysis and neural network-based modeling of increasing thickness PMMA sheets as cut by CO2 laser

Abstract

One crucial issue in laser beam cutting is the proper selection of processing factors to achieve straight kerfs. In this study, the influence of cutting speed, laser power, gas pressure, and focal point position on the CO2 laser cutting of polymethylmethacrylate (PMMA) sheets of different thicknesses is analyzed to select optimal operational parameters. A design of experiment (DOE) of response surface type was then carried out to identify the influence of each factor and set the optimal ranges of the process parameters to achieve flat kerfs without streaks or surface imperfections. A laser with power values ranging from 3000 to 4200 W was used as an energy source. The interest intervals for the rest of the process parameters were justified by the literature consulted. Three new criteria were established to estimate a kerf profile deviation (KPD) response considering all the kerf area cut by the laser, defining these criteria as the difference between the resulting kerf profile and a straight kerf. As final results, the kerf profile deviation response was improved by 22.6%, 42.4%, 15.6%, and 22.1% in PMMA having thicknesses of 4, 8, 12, and 20 mm, respectively. Finally, an artificial neural network (ANN)-based model is used to predict the value of KPD, taking into account all the variables involved in the cutting process and the different criteria applied to classify the stage of profile deviation. The contributions of this paper are, in the first place, the proposal of new metrics to evaluate the kerf profile in PMMA sheets. Secondly, establishing optimal cutting parameters and ANN as an accurate method in geometrical modeling and relating the influential variables on the laser cutting process.