Search
Menu

Spectroscopy, Machine Learning Pair to Ease Recyclables Sorting

Facebook X LinkedIn Email
Researchers at Hefei University of Technology have demonstrated the use of laser-induced breakdown spectroscopy (LIBS) to automatically identify, classify, and subclassify recyclable waste in real time. The resource re-use application enabled the researchers to identify and sort samples, based on material composition, into six consumer-level categories: paper, plastic, glass, metal, textile, and wood.  

The detection method holds promise as a solution for the environmental protection and waste management fields, where improved solutions to identifying and automatically classifying recyclable waste are essential to environmental sustainability.

To maximize the re-use of material, waste products must be disposed of based on their material properties. Existing methods to identify and categorize waste based on physical characteristics, or by using image-based techniques, can be inaccurate and unreliable. Current spectroscopy methods for identifying and classifying waste must process the detected samples in advance, which rules out automatic, real-time detection.

The researchers combined LIBS technology with drop-dimension algorithms and machine-learning algorithms. In addition to detecting the elemental composition of a sample based on emission spectra, LIBS is not affected by the ambient environment and light, or the shape and color of the sample. Further, the technique does not require preprocessing of the sample. 

The researchers collected the spectra of 80 recyclable waste samples, using LIBS to make the initial set of identifications and classifications. They next used LIBS to subclassify metals and plastics for reuse at the level of a recycling factory. Metals were subclassified into the subcategories of iron, stainless steel, copper, and aluminum. Plastics were subclassified into six subcategories.


Due to the high dimension and large amount of redundant information in the LIBS spectral data, the researchers applied principal component analysis and linear discriminant analysis to drop the dimensions of the collected full-spectra data. They input the drop-dimensional spectra into random forest and back propagation neural network (BPNN) machine-learning models for training models and predicting results.

Among those that the researchers explored, the optimal model for classifying recyclable waste and the most effective model for subclassifying metals and plastics were determined. They achieved 100% accuracy in classifying recyclable waste. They achieved 98.77% accuracy for metal subclassification and 99.52% accuracy for plastic subclassification.  
The identification and classification system for recyclable waste. Courtesy of Lei Yang.

 


The identification and classification system for recyclable waste. Courtesy of Lei Yang.
The researchers plan to increase the number of waste samples in their study and incorporate other forms of waste, such as kitchen waste, into their investigations. They also hope to use LIBS to expand scientific understanding of transparent glass detection, which could open new avenues for recycling and waste management.

The research was published in AIP Advances (www.doi.org/10.1063/5.0149329).

Published: August 2023
Glossary
laser-induced breakdown spectroscopy
Laser-induced breakdown spectroscopy (LIBS) is an analytical technique that uses a high-powered laser pulse to ablate a small amount of material from a sample, creating a plasma. This plasma emits light, which is analyzed to determine the elemental composition of the sample. Principle of operation: A focused laser pulse is directed at the sample, causing rapid heating and vaporization of a small amount of material. The vaporized material forms a high-temperature plasma, which consists of...
machine learning
Machine learning (ML) is a subset of artificial intelligence (AI) that focuses on the development of algorithms and statistical models that enable computers to improve their performance on a specific task through experience or training. Instead of being explicitly programmed to perform a task, a machine learning system learns from data and examples. The primary goal of machine learning is to develop models that can generalize patterns from data and make predictions or decisions without being...
Research & TechnologyeducationAsia-PacificHefei University of TechnologyLasersspectroscopylaser-induced breakdown spectroscopysustainabilityConsumerindustrialenvironmentrecyclingwaste managementmachine learningneural networksMaterialsSensors & DetectorsautomationLight SourcesTechnology News

We use cookies to improve user experience and analyze our website traffic as stated in our Privacy Policy. By using this website, you agree to the use of cookies unless you have disabled them.
BioPhotonics Conference 2024LIVE NOW: Label-Free Multiphoton Microcopy Advances Clinical Skin Imaging X