Evaluation of Soft Computing Methods for Parameters Estimation and Sensitivity Analysis of Laser Cutting
M. Milovančević, H. Deneva, L. Lazov, V. Nikolić and D. Petković
Laser cutting process with water jet assistance could produces better performances than traditional laser cutting process. Since it is highly nonlinear process there is need to predict and to estimate the most influential factors for the process in order to increase the efficiency of the process. So in this study parameters estimation of the laser cutting process with water jet was performed. To make the easy application of the process one needs to know in ahead which parameters are to most influential to the process and what will be the output of the laser cutting process with water jet assistance. Because of high nonlinearly of the process, soft computing algorithm was used. Three algorithms were implemented, support vector regression (SVR), artificial neural network (ANN) and genetic programming (GP). SVR method was shown the best prediction performances of the parameters of the laser cutting process with water jet assistance. The best forecasting accuracy was observed for prediction of striation depth (R2 = 0.9987). The worst forecasting accuracy was observed for prediction of top kerf width (R2 = 0.9814). Sensitivity analysis was shown that the laser power has the highest influence on the mostly parameters. Cutting speed has the most influence on the top kerf width.
Keywords: Laser cutting, water jet, soft computing, algorithm, support vector regression (SVR), artificial neural network (ANN), genetic programming (GP), forecasting