Related Experiment Video
Updated: May 22, 2025
06:50
O-cresol Concentration Online Measurement Based On Near Infrared Spectroscopy Via Partial Least Square Regression
Published on: November 8, 2019
6.5K
Identification of Aged Polypropylene with Machine Learning and Near-Infrared Spectroscopy for Improved Recycling.
Keyu Zhu1,2,3, Delong Wu1,2,3, Songwei Yang1,2,3
1College of Environmental and Resource Sciences, College of Carbon Neutral Modern Industry, Fujian Normal University, Fuzhou 350007, China.
Polymers
|March 13, 2025
Summary
This study uses machine learning and near-infrared (NIR) spectroscopy to accurately identify aged polypropylene (PP) plastics. This advances automated, sustainable plastic recycling by improving sorting efficiency and material quality.
Area of Science:
- Materials Science
- Analytical Chemistry
- Computer Science
Background:
- Traditional plastic sorting is manual, inefficient, and unsafe.
- Near-infrared (NIR) spectroscopy offers rapid, non-destructive analysis but faces challenges with complex spectral data.
- Distinguishing aged plastics is crucial for effective recycling.
Purpose of the Study:
- To develop a machine learning model using NIR spectroscopy to classify polypropylene (PP) at various aging stages.
- To evaluate the impact of aging on PP mechanical properties.
- To optimize spectral preprocessing techniques for enhanced classification accuracy.
Main Methods:
- Collected NIR spectra from PP samples aged under simulated conditions.
- Utilized Fourier-transform infrared spectroscopy (FTIR) to confirm aging.
- Performed mechanical property tests (tensile strength, elongation at break).
- Developed and evaluated machine learning classification models, including spectral preprocessing techniques like the second derivative method with linear Support Vector Classification (SVC).
Main Results:
- Mechanical properties, particularly elongation at break, significantly decreased with aging.
- The optimized NIR spectroscopy and machine learning model achieved 99% classification accuracy and 100% precision in identifying PP aging stages.
- The second derivative method combined with linear SVC proved highly effective.
Conclusions:
- Accurate identification of PP at different aging stages is feasible using machine learning-based NIR spectroscopy.
- This approach enhances the quality and efficiency of recycled plastics.
- The study promotes automated, precise, and sustainable plastic recycling processes.
