Optimization of the Parameters Technological Operation of Laser Cutting of Silicate Glass Using a Genetic Algorithm and Neural Networks

Abstract

This paper presents the process of numerical modeling and optimization of the dual-beam laser cleaving process for silicate glasses. Technological parameters enabling effective separation of glass plates under the action of two laser beams are identified. Temperature fields and thermoelastic stresses were computed using the finite element method in a quasi-static formulation, implemented in Python with the FEniCS library. Process optimization was carried out using a modified genetic algorithm, with the objectives of maximizing tensile stress and processing speed. The varied parameters included laser power, processing speed, and laser spot radius (wavelength 10.6 μm). The responses considered were the maximum temperatures and stresses within the laser-affected zone. A regression model of the process was developed. The error of the results when using a neural network approximation did not exceed 4% for temperature and 5% for stress. The paper also describes a real-time adaptive control approach based on neuroregulators, which ensures high precision and stability of the processing operation.