Related Experiment Video
Updated: Aug 7, 2025
10:16
Protocol for Microplastics Sampling on the Sea Surface and Sample Analysis
Published on: December 16, 2016
49.8K
Classification of household microplastics using a multi-model approach based on Raman spectroscopy
Zikang Feng1, Lina Zheng2, Jia Liu1
1School of Safety Engineering, China University of Mining and Technology, Xuzhou, People's Republic of China.
Chemosphere
|March 12, 2023
Summary
Identifying household microplastics is challenging. A multi-model machine learning approach combined with Raman spectroscopy accurately classifies microplastics, achieving over 98% accuracy for various sample types.
Area of Science:
- Environmental Science
- Analytical Chemistry
- Materials Science
Background:
- Microplastic pollution is a significant environmental concern, originating from the widespread use of plastics.
- Household plastic products contribute substantially to microplastic release, posing challenges due to their small size and complex composition.
- Accurate identification and quantification of microplastics are crucial for understanding their environmental impact.
Purpose of the Study:
- To develop and validate a robust machine learning approach for the classification of household microplastics.
- To enhance the accuracy of microplastic identification using Raman spectroscopy combined with advanced algorithms.
- To assess the performance of the developed models on standard, real-world, and environmentally stressed microplastic samples.
Main Methods:
- Utilized Raman spectroscopy for microplastic sample analysis.
- Implemented and compared four single machine learning models: Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Linear Discriminant Analysis (LDA), and Multi-layer Perceptron (MLP).
- Employed Principal Component Analysis (PCA) for dimensionality reduction prior to SVM, KNN, and LDA, and the reliefF algorithm for distinguishing specific plastic types (HDPE, LDPE). Developed a multi-model approach integrating PCA-LDA, PCA-KNN, and MLP.
Main Results:
- Single machine learning models achieved over 88% classification accuracy for standard microplastic samples.
- The multi-model approach demonstrated superior performance, reaching over 98% recognition accuracy for standard, real, and environmentally stressed microplastic samples.
- The reliefF algorithm effectively differentiated between High-Density Polyethylene (HDPE) and Low-Density Polyethylene (LDPE) samples.
Conclusions:
- The combination of Raman spectroscopy and a multi-model machine learning approach provides a highly accurate and valuable tool for microplastic classification.
- This method offers a reliable solution for identifying and quantifying microplastics from diverse sources, including household products.
- The developed approach has significant potential for environmental monitoring and plastic pollution research.
