Prediction of Femtosecond Laser Etching Parameters Based on a Backpropagation Neural Network with Grey Wolf...

Area of Science:

• Materials Science and Engineering
• Artificial Intelligence
• Manufacturing Technology

Background:

• Optimizing laser processing parameters for industrial applications is complex and time-consuming.
• Existing analytical models for laser processing are limited due to intricate mechanisms.

Purpose of the Study:

• To develop a predictive model for laser etching parameters and surface characteristics.
• To reduce the time and energy required for laser process optimization using AI.

Main Methods:

• Developed a backpropagation neural network (BPNN) integrated with a grey wolf optimization (GWO) algorithm.
• Utilized the Keras API in Python for model development.
• Trained and validated the GWO-BPNN model on experimental data from a 30 W laser source.

Main Results:

• The GWO-BPNN model accurately predicted multi-input laser etching parameters (energy, scanning velocity, number of exposures).
• The model precisely forecasted multi-output surface characteristics (depth, width).
• Achieved an R-squared (R²) score exceeding 0.90, indicating excellent prediction accuracy.

Conclusions:

• The developed GWO-BPNN model offers a fast and accurate solution for predicting laser etching outcomes.
• This AI-driven approach assists engineers in optimizing laser processing, reducing experimental effort and resource consumption.