Convolution Neural Network with Laser-Induced Breakdown Spectroscopy as a Monitoring Tool for Laser Cleaning Process

Area of Science:

• Materials Science and Engineering
• Analytical Chemistry
• Artificial Intelligence in Manufacturing

Background:

• Laser cleaning is an effective method for removing paint and coatings from stainless steel.
• Monitoring the effectiveness of laser cleaning in real-time is crucial for process optimization.
• Analyzing Laser-Induced Breakdown Spectroscopy (LIBS) data for precise material identification can be complex, especially with similar chemical compositions.

Purpose of the Study:

• To develop an automated and accurate method for analyzing LIBS spectra obtained during laser cleaning of painted stainless steel.
• To investigate the application of deep learning, specifically convolutional neural networks (CNN), for rapid LIBS data interpretation.
• To assess the feasibility of integrating a LIBS-coupled CNN system as an in-line monitoring tool for laser cleaning processes.

Main Methods:

• Eight painted stainless steel 304L specimens were subjected to laser cleaning with varied parameters (laser power, scan speed, repetitions).
• Laser-Induced Breakdown Spectroscopy (LIBS) was employed to collect spectral data during the cleaning process.
• A convolutional neural network (CNN)-based deep learning model was developed and trained for the classification and analysis of LIBS spectra.

Main Results:

• The developed CNN model achieved a classification accuracy of 94.55% for identifying laser-cleaned stainless steel specimens.
• The analysis time for each LIBS spectrum using the CNN method was significantly reduced to 0.09 seconds.
• The study demonstrated the effectiveness of the LIBS-coupled CNN approach in distinguishing between cleaned and uncleaned states.

Conclusions:

• The LIBS-coupled CNN method provides an accurate and rapid solution for analyzing spectral data during laser cleaning.
• This deep learning approach overcomes the challenges of identifying LIBS spectra with similar chemical compositions.
• The findings support the potential of using this integrated system as a viable in-line tool for real-time monitoring and control of laser cleaning operations.