基于AFSA-SVM动态光谱的血液识别研究  

Blood Identification Based on AFSA-SVM Dynamic Spectra

作　　者：马焕臻 闫薪如 辛英健 方沛沛 王泓鹏 王一安 段明康 贾建军[3] 何继业[2] 万雄 MA Huan-zhen;YAN Xin-ru;XIN Ying-jian;FANG Pei-pei;WANG Hong-peng;WANG Yi-an;DUAN Ming-kang;JIA Jian-jun;HE Ji-ye;WAN Xiong(Key Laboratory of Systems Health Science of Zhejiang Province,School of Life Science,Hangzhou Institute for Advanced Study,University of Chinese Academy of Sciences,Hangzhou 310024,China;Department of Orthopedics,Xinhua Hospital,Shanghai Jiao Tong University School of Medicine,Shanghai 200092,China;Key Laboratory of Space Active Opto-Electronics Technology of the Chinese Academy of Sciences,Shanghai Institute of Technical Physics,Chinese Academy of Sciences,Shanghai 200083,China;University of the Chinese Academy of Sciences,Beijing 100049,China)

机构地区：[1]浙江省系统健康科学重点实验室,国科大杭州高等研究院生命与健康科学学院,浙江杭州310024 [2]上海交通大学医学院附属新华医院骨科,上海200092 [3]中国科学院空间主动光电技术重点实验室,中国科学院上海技术物理研究所,上海200083 [4]中国科学院大学,北京100049

出　　处：《光谱学与光谱分析》2024年第7期1877-1882,共6页Spectroscopy and Spectral Analysis

基　　金：国家重点科研计划项目(2021YFF0601201,2021YFA0716100,2018YFC1200202);国家自然科学基金项目(42074210);上海市自然科学基金项目(21ZR1473700,19ZR1465800);上海技术物理研究所创新专项(CX310,CX364);上海市科技重大项目(2019SHZDZX01);上海市基础研究特区计划项目(JCYJ-SHFY-2021-04)资助。

摘　　要：血液是一种受管制的特殊遗传生物资源。针对传统血液光谱检测中易氧化变质的问题,采用基于仿生血管的动态共聚焦拉曼荧光光谱,开展了猪、马、鸽、鸡、鸭、鹅等六种家禽家畜的血液物种鉴别研究。原始光谱的预处理过程包括去基线、平滑和归一化等。采用线性判别分析对光谱数据进行降维处理,继而用支持向量机建立识别模型,选用高斯核函数,通过人工鱼群算法优化支持向量机的参数C和γ,使其分类准确率最高,最优的C和γ分别为0.2和0.134。人工鱼群-支持向量机模型识别准确率达到97.2%,基于仿生血管的动态共聚焦拉曼荧光光谱可以满足血液安全高效的检测要求,用人工鱼群算法优化支持向量机参数的算法模型表现出较好的分类效果。Blood is a regulated exceptional genetic biological resource.In response to the issue of easy oxidation and deterioration in traditional blood spectral detection,dynamic confocal Raman fluorescence spectroscopy technology based on biomimetic blood vessels was used to conduct blood species identification research on six types of poultry and livestock,including pigs,horses,pigeons,chickens,ducks,and geese.The preprocessing process of the original spectrum includes baseline removal,smoothing,and normalization.Linear discriminant analysis is used to reduce the dimensionality of spectral data,and then support vector machines are used to establish recognition models.Gaussian kernel functions are selected,and the parameters C and γ Make their classification accuracy the highest,the optimal C and γ 0.2 and 0.134,respectively.The recognition accuracy of the artificial fish school support vector machine model reaches 97.2%.The dynamic confocal Raman fluorescence spectrum based on biomimetic blood vessels used in this article can meet the requirements of blood safety and efficiency detection,and the algorithm model optimized by the artificial fish school algorithm for support vector machine parameters shows good classification performance.

关 键 词：人工鱼群算法 共聚焦拉曼光谱 支持向量机 

分 类 号：O561.3[理学—原子与分子物理]