研究目的
Investigating the application of LIBS technique combined with machine learning algorithms for the discrimination/identification of different plastic/polymeric samples containing various additives.
研究成果
The combination of LIBS technique with machine learning algorithms (PCA and LDA) achieved identification accuracies as high as 100% for the classification of plastic samples based on their additives. This approach is efficient for sorting and recycling purposes.
研究不足
The study focuses on ABS polymers with specific additives. The effectiveness of LIBS for detecting bromine in brominated flame retardants is limited due to the spectral emissions lying in the VUV or NIR regions.
1：Experimental Design and Method Selection
LIBS technique was used for the discrimination/identification of plastic/polymeric samples. Machine learning algorithms (PCA and LDA) were employed for the analysis of the LIBS spectroscopic data.
2：Sample Selection and Data Sources
Industrial grade Acrylonitrile Butadiene Styrene polymers (ABS) with different additives were used. Data were collected from LIBS spectra extended from 200 to 1100 nm.
3：List of Experimental Equipment and Materials
5 ns Q-Switched Nd:YAG laser, portable spectrograph (Avantes AvaSpec-2048-USB2), monochromator (Jobin–Yvon HR460), ICCD camera (Andor iStar DH712), quartz lenses, optical fiber bundles.
4：Experimental Procedures and Operational Workflow
Laser beam focused on the sample surface to induce plasma. Emission spectra collected and analyzed. PCA and LDA algorithms were trained using the in-built routines in the Scikit-learn library.
5：Data Analysis Methods
PCA for dimensionality reduction and visualization. LDA for classification. 10-fold cross-validation technique for accuracy evaluation.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
