Optimization of CO2 Laser Cutting Parameters Using Adaptive Neuro-Fuzzy Inference System (ANFIS)

Abstract

Laser cutting is a manufacturing technology that uses laser light to cut almost any materials. This type of cutting technology has been applied in many industrial applications. Problems seen with a laser is the cutting efficiency and the quality wherein these two parameters are both affected by the laser power and its process speed. This study presents the modelling and simulation of an intelligent system for predicting and optimising the process parameters of CO2 laser cutting. The developed model was trained and tested using actual data gathered from actual laser cut runs. For the system parameters, two inputs were used: the type of material used and the material thickness (mm). For the desired response, the output is the process speed or cutting rate (mm/min). Adaptive neuro-fuzzy inference system (ANFIS) was the tool used to model the optimisation cutting process. Moreover, grid partition (GP) and subtractive clustering were both used in designing the fuzzy inference system (FIS). Among the training models used, GP Gaussian bell membership function (Gbellmf) provided the highest performance with an accuracy of 99.66%.