支持向量机结合主成分分析辅助激光诱导击穿光谱技术识别鲜肉品种

发布时间：2018-06-10 11:21

本文选题：激光诱导击穿光谱 + 支持向量机　；参考：《分析化学》2017年03期

【摘要】：为提高激光诱导击穿光谱技术(Laser-induced breakdown spectroscopy,LIBS)对鲜肉品种的识别率,采用支持向量机结合主成分分析算法辅助LIBS技术对鲜肉品种进行识别。对鲜肉切片用载玻片压平,采用LIBS技术对鲜肉组织(猪肉、牛肉和鸡肉)表面进行光谱数据的采集,每种鲜肉采集150幅光谱并进行随机排列,取前75幅光谱作为训练集建立模型,后75幅作为测试集测试建模结果。研究选取K、Ca、Na、Mg、Al、H、O等元素的49条归一化谱线数据进行主成分分析,并用所得数据建立支持向量机分类模型。结果表明,通过主成分分析降维,输入变量从49个优化减少到18个,模型建模速度从88.91 s降至55.52 s,提高了支持向量机的建模效率;并使预测集的平均识别率提高到89.11%。本研究为激光诱导击穿光谱技术在鲜肉品种快速分类领域提供了方法和数据参考。
[Abstract]:In order to improve the recognition rate of fresh meat varieties by Laser-induced breakdown spectroscopy (LIBS), support vector machine (SVM) combined with principal component analysis (PCA) algorithm was used to identify fresh meat varieties. The surface of fresh meat tissue (pork, beef and chicken) was collected by Libs technique. 150 spectra of each meat were collected and arranged randomly. The first 75 spectra were used as the training set to establish the model, and the latter 75 as the test set test modeling results. The principal component analysis (PCA) of 49 normalized spectral line data of elements such as KKCa-Ca-NaMg-Mg-AL-HZO was carried out, and the classification model of support vector machine was established with the obtained data. The results show that the input variables are reduced from 49 optimizations to 18 and the modeling speed is reduced from 88.91 s to 55.52 s through principal component analysis (PCA), which improves the modeling efficiency of SVM, and increases the average recognition rate of prediction set to 89.11 s. This study provides a method and data reference for fast classification of fresh meat varieties 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)激光与太赫兹技术功能实验室;
【基金】：国家重大科学仪器设备开发专项(No.2011YQ160017) 国家自然科学基金项目(No.6157031235)资助~~
【分类号】：TS251.7;O657.3

【相似文献】