塑料激光诱导击穿光谱技术快速分类应用研究

发布时间：2018-06-01 02:15

本文选题：激光诱导击穿光谱技术 + 非金属元素特征谱线　；参考：《光谱学与光谱分析》2017年07期

【摘要】：在大气环境中,采用激光诱导击穿光谱技术与支持向量机算法相结合,对来自不同厂家不同颜色的20种工业塑料进行分类研究。首先对分类结果有影响的实验参数进行优化,在最佳的实验参数条件下进行光谱采集,采用6条非金属元素特征谱线,有效缩短了训练支持向量机分类模型所需时间,从而提高了塑料的分类效率。实验结果表明,利用碳、氢、氧、氮等主量非金属元素对这些工业塑料样品进行分类,测试集1 000个光谱数据中有996个识别正确,算术平均识别精度达到99.6%。在选取较少的主量非金属特征谱线的情况下,结合采用支持向量机算法,可以实现激光诱导击穿光谱技术对更多类型的塑料制品快速、高精度分类,为激光诱导击穿光谱技术在实现塑料分类方面提供了数据参考。
[Abstract]:In the atmospheric environment, 20 kinds of industrial plastics from different manufacturers and different colors were classified and studied by combining laser induced breakdown spectroscopy with support vector machine (SVM). Firstly, the experimental parameters which have an effect on the classification results are optimized, and the spectral collection is carried out under the optimal experimental parameters, and six characteristic spectral lines of non-metallic elements are adopted, which effectively shortens the time required for training the classification model of support vector machine (SVM). Thus, the classification efficiency of plastics is improved. The experimental results show that 996 of the 1000 spectral data are correctly identified and the arithmetic average recognition accuracy is 99.6. The industrial plastics samples are classified by carbon, hydrogen, oxygen, nitrogen and other major nonmetallic elements. In the case of selecting fewer principal nonmetallic characteristic lines and combining with support vector machine (SVM) algorithm, laser induced breakdown spectroscopy can be used to classify more plastic products quickly and accurately. It provides a data reference for the classification of plastics by laser induced breakdown spectroscopy.
【作者单位】：华中科技大学武汉光电国家实验室(筹)激光与太赫兹功能实验室;
【基金】：中央高校基本科研业务费专项资金项目(2014QNRCO24,2015MS002)资助
【分类号】：O657.319;TQ320.77

【相似文献】