Citation: Choi, S.; Park, C.
Convolution Neural Network with
Laser-Induced Breakdown
Spectroscopy as a Monitoring Tool
for Laser Cleaning Process. Sensors
2023, 23, 83. https://doi.org/
10.3390/s23010083
Academic Editor: Xinyu Li
Received: 23 November 2022
Revised: 18 December 2022
Accepted: 19 December 2022
Published: 22 December 2022
Copyright: © 2022 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Article
Convolution Neural Network with Laser-Induced Breakdown
Spectroscopy as a Monitoring Tool for Laser Cleaning Process
Soojin Choi
1
and Changkyoo Park
1,2,
*
1
Department of Laser and Electron Beam Technologies, Korea Institute of Machinery and Materials,
Daejeon 34103, Republic of Korea
2
Department of Materials Science and Engineering, Seoul National University of Science and Technology,
Seoul 01811, Republic of Korea
* Correspondence: ck0421@seoultech.ac.kr
Abstract:
In this study, eight different painted stainless steel 304L specimens were laser-cleaned
using different process parameters, such as laser power, scan speed, and the number of repetitions.
Laser-induced breakdown spectroscopy (LIBS) was adopted as the monitoring tool for laser cleaning.
Identification of LIBS spectra with similar chemical compositions is challenging. A convolutional
neural network (CNN)-based deep learning method was developed for accurate and rapid analysis
of LIBS spectra. By applying the LIBS-coupled CNN method, the classification CNN model accuracy
of laser-cleaned specimens was 94.55%. Moreover, the LIBS spectrum analysis time was 0.09 s. The
results verified the possibility of using the LIBS-coupled CNN method as an in-line tool for the laser
cleaning process.
Keywords:
laser cleaning; paint removal; monitoring; laser-induced breakdown spectroscopy;
convolution neural network
1. Introduction
Laser cleaning is the technique to remove contaminants from surfaces by laser ablation,
which occurs when a high-energy laser pulse irradiates the sample surface [
1
]. Laser
ablation removes contaminants via the materials’ evaporation and volatilization. Laser
cleaning has been widely used in art restorations [
2
,
3
], paint removal [
4
], and maintenance
of metal alloys [
5
]. Monitoring techniques have been applied for overcleaning prevention
and residual contaminant detection for laser cleaning. Acoustic signal [
4
] and FE-SEM
and EPMA [
6
] were used to monitor natural marine microbiofoulings and paint removal
on metal surface, respectively. Micro-CT and micro-XRF [
7
] were adopted to examine the
laser cleaning level of black crusts on limestone monuments. However, those techniques
may not be suitable as an in-line monitoring for the laser cleaning process. This is because
the high-power laser cleaning process produces extremely large noise. In addition, the
above-mentioned techniques (except acoustic signal) require thorough pretreatments to
obtain accurate analysis results.
Laser-induced breakdown spectroscopy (LIBS) is an effective technique as an in-line
monitoring for the laser cleaning process by investigating the elemental composition using
laser-induced plasma. Rapid spectral data analysis allows LIBS to be applied for an in-
line monitoring. However, identifying different types of samples with similar elemental
compositions using LIBS analysis is challenging. Therefore, machine learning and deep
learning methods are being adopted to improve the accuracy and speed of LIBS analyses.
Sirven et al. [
8
] adopted principal component analysis (PCA), soft independent modeling
of class analogy (SIMCA), and partial least-squares discriminant analysis (PLS-DA) for
rock classification. Yelameli et al. [
9
] used a support vector machine (SVM) to distinguish
ten different rock samples. Li et al. [
10
] applied k-nearest neighbors (kNN) and an SVM to
discriminate soft tissues.
Sensors 2023, 23, 83. https://doi.org/10.3390/s23010083 https://www.mdpi.com/journal/sensors
