支持向量机算法在激光诱导击穿光谱技术塑料识别中的应用研究

基于支持向量机 (support vector machines, SVM) 算法采用激光诱导击穿光谱技术对11种塑料进行了识别. 每种塑料各采集100个光谱, 其中50个光谱作为训练集, 用于建立支持向量机模型, 剩下的50 个光谱作为测试集, 用于测试所建立支持向量机模型的识别精度. 结果表明测试集550个光谱中有543个光谱识别正确,算术平均识别精度达到了98.73%. 其中有6个聚氨酯 (PU) 光谱被误判为有机玻璃 (PMMA), 原因主要是受空气中氮气的影响, 使得有机玻璃和聚氨酯两种塑料在氮元素含量上的差异不能通过N I 746.87 nm, C-N(0,0) 388.3 nm两条谱线的强度准确表征. 本结果为LIBS技术塑料分类提供了方法和数据参考. 关键词： 支持向量机 激光诱导击穿光谱 塑料识别

Identification of plastics by laser-induced breakdown spectroscopy combined with support vector machine algorithm

Laser-induced breakdown spectroscopy (LIBS) combined with support vector machine (SVM) algorithm was used to identify 11 kinds of plastics. For each plastic, 100 spectra recorded by the spectrometer system were divided equally into training set and test set, and the former was used to train SVM model while the latter was used to validate SVM model created by the training set. Result shows that 543 of 550 test set spectra are identified correctly with the average correct identification rate 98.73%. However, there are six spectra of PU misidentified as PMMA. This is because the difference of nitrogen content in 11 plastics cannot be reflected by the intensities of N I 746.87 nm and C-N (0,0) 388.3 nm due to the influence of ambient air. Methods and reference data are provided for further study of plastics identification by laser-induced breakdown spectroscopy technique.
Keywords： laser-induced breakdown spectroscopy support vector machine plastics identification