Multi-objective optimization of fiber laser cutting quality characteristics of glass fiber reinforced plastic (GFRP) materials

Abstract

Laser cutting can achieve ultra-precision cutting of materials by using appropriate parameters. In order to optimize processing parameters and obtain better processing quality, conducted experimental research on laser cutting of Glass fiber reinforced plastic (GFRP). In this study, the impact of four process parameters (laser power, cutting speed, assistant gas pressure and focus position) on quality characteristics (kerf width, kerf taper and kerf section roughness) was evaluated through analysis of variance (ANOVA), the regression relationship between laser processing parameters and quality characteristics was analyzed. And an integrative model for predicting and optimizing the quality characteristics of fiber laser cutting was constructed. In the proposed integrative model, Back-Propagation Neural Network (BPNN) was used to build a prediction model for quality characteristics. Through Non-dominated Sorting Genetic Algorithm (NSGAII) optimizes and outputs the complete optimal solution set of processing parameters, and finally realizes the nonlinear optimization of multi-objective parameters. The fitness value of BPNN model can reach 97.814%, and the maximum prediction relative error is 8.614%. And the set of suggested optimal solutions can be used as most of them are showing certain improvements in the quality characteristics. The results indicate that the integrative modeling strategy has the ability to predict and optimize laser cutting of composite materials.