Feasibility Study on Identification of Mercury Element From An Unknown Substance Using Visible-NIR Spectroscopy and SVM Classifier
碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 106 === The identification of elements inside materials is very essential and important part of spectroscopy. Support Vector Machine(SVM) has proven to be powerful in spectroscopy. Support vector based identification of elements from its spectrum is proposed in thi...
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Others |
| Language: | en_US |
| Published: |
2018
|
| Online Access: | http://ndltd.ncl.edu.tw/handle/xytxr5 |
| Summary: | 碩士 === 國立臺灣科技大學 === 自動化及控制研究所 === 106 === The identification of elements inside materials is very essential and important part of spectroscopy. Support Vector Machine(SVM) has proven to be powerful in spectroscopy. Support vector based identification of elements from its spectrum is proposed in this study. Here in this study HG-AR(Mercury-Argon) samples are collected from Ocean Optics Spectrometer HR4000 in order to identify Mercury from any unknown substance. Spectral features are extracted from raw data in terms of peak intensity and their wave-length to differentiate two categories.Support vector machine as binary classifier is used to identify element from their spectral features. The identification results are achieved from Nu-SVM and Polynomial SVM classifier in under different parameters. The best identification results were achieved by Nu-SVM classifier. The identification accuracy was 100\textdiscount~ at nu=0.1,0.2 and polynimial degree=2 and 3. The overall results fortify that spectroscopy with SVM can be coherent and rapid to identify elements from any unknown substances.
|
|---|