New Model for Scrap Metal Identification Using Laser-Induced Breakdown Spectroscopy and Machine Learning

Researchers at the Gwangju Institute of Science and Technology (GIST) in South Korea have developed a new model for identifying scrap metal using laser-induced breakdown spectroscopy (LIBS) combined with machine learning (1). The proposed model uses a two-step process called Aug2Tran, which includes the augmentation of the standard reference material (SRM) data set and the use of transfer learning to build a robust real-time classification model.

Scrap metal background. | Image Credit: © uwimages - stock.adobe.com

The two-step Aug2Tran model is a transfer learning-based classification model for identifying scrap metal using an augmented training data set consisting of LIBS measurements of SRM samples. The first step involves augmenting the SRM data set by synthesizing spectra of unobserved types through attenuation of dominant peaks corresponding to sample composition and generating spectra depending on the target sample using a generative adversarial network. The second step uses the augmented SRM data set to build a robust real-time classification model with a convolutional neural network, which is further customized for the target scrap metal with limited measurements through transfer learning. This approach improves the classification accuracy of arbitrarily shaped static or moving samples with various surface contaminations and compositions, and even for differing ranges of expected intensities and wavelengths. The proposed Aug2Tran model can be used as a systematic model for scrap metal classification with generalizability and ease of implementation.

The researchers addressed the challenges of a limited training set and differences in experimental configuration by synthesizing spectra of unobserved types through attenuation of dominant peaks and generating spectra using a generative adversarial network. They used the augmented SRM data set to build a convolutional neural network, which was further customized for the target scrap metal through transfer learning.

The model was evaluated by measuring SRMs of five representative metal types and testing with scrap metal from actual industrial fields under three different configurations. The proposed scheme produced an average classification accuracy of 98.25%, as high as the results of the conventional scheme with three separately trained and executed models.

This new model improves the classification accuracy of arbitrarily shaped static or moving samples with various surface contaminations and compositions, and even for differing ranges of charted intensities and wavelengths. The Aug2Tran model can be used as a systematic model for scrap metal classification with generalizability and ease of implementation.

LIBS provides a unique and fast way of identifying unknown samples without complicated sample preparation. Combining LIBS with machine learning has been actively studied for industrial applications such as scrap metal recycling. The proposed Aug2Tran model provides an efficient and accurate way of identifying scrap metal, which can contribute to the efficient and sustainable use of resources. The findings of this study are published in the journal Applied Spectroscopy (1).

Reference

(1) Srivastava, E.; Kim, H.; Lee, J.; Shin, S.; Jeong, S.; Hwang, E. Adversarial Data Augmentation and Transfer Net for Scrap Metal Identification Using Laser-Induced Breakdown Spectroscopy Measurement of Standard Reference Materials. Appl. Spectrosc. 2023, ASAP. DOI: 10.1177/00037028231170234