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  

Classification of Fresh Meat Species Using Laser-induced Breakdown Spectroscopy with Support Vector Machine and Principal Component Analysis

作　　者：朱毅宁 杨平[1] 杨新艳[1] 李嘉铭[1] 郝中骐[1] 李秋实[1] 郭连波[1] 李祥友[1] 曾晓雁[1] 陆永枫[1] 

机构地区：[1]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,武汉430074

出　　处：《分析化学》2017年第3期336-341,共6页Chinese Journal of Analytical Chemistry

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~~

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。To improve the classification accuracy of fresh meat species using laser-induced breakdown spectroscopy （LIBS）, the support vector machine （SVM） and principal component analysis （PCA） were combined to classify fresh meat species （including pork, beef, and chicken）. A simple sample preparation to flatten fresh meat by glass slides was proposed. For each meat sample, 150 spectra were recorded and randomly arranged. The first 75 spectra were used to train a model while the others were used for model validation. By analyzing the 49 normalized spectral lines （K, Ca, Na, Mg, AI, H, O, etc. ） in the different tissues, the classification model was built. The results showed that the dimensionality of input variables was decreased from 49 to 10 and modeling time was reduced from 89.11 s to 55.52 s using PCA, thus improving the modeling efficiency. The mean classification accuracy of 89.11% was achieved. The method and reference data are provided for further study of fresh meat classification by laser-induced breakdown spectroscopy technique.

关 键 词：激光诱导击穿光谱 支持向量机 主成分分析 组织分类 

分 类 号：TS251.7[轻工技术与工程—农产品加工及贮藏工程] O657.3[轻工技术与工程—食品科学与工程]