【作者】 吴海云； 卫勇； 杨仁杰； 单慧勇； 曾雅楠； 董桂梅； 艾成龙；

【Author】 WU Hai-yun;WEI Yong;YANG Ren-jie;SHAN Hui-yong;ZENG Ya-nan;DONG Gui-mei;AI Cheng-long;College of Engineering and Technology,Tianjin Agricultural University;Sinotech(Tianjin) Intelligent System Engineering Co.,Ltd.;

【机构】 天津农学院工程技术学院； 中科国技(天津)智能系统工程有限责任公司；

【摘要】 奶牛乳房炎是危害世界奶牛业的主要因素之一。研究提出一种基于二维近红外相关谱结合模式识别的奶牛乳房炎快速鉴别方法。首先以不同患病程度奶牛的乳汁为研究对象,在350～2 500 nm范围内采集并分析了40个黑白花奶牛乳汁样品光谱。结果发现健康和患病奶牛乳汁的常规一维近红外光谱图近似,进一步以患病程度为外绕,进行二维同步相关光谱分析。建立了基于二维相关谱特征矩阵的多维偏最小二乘判别模型(NPLS-DA),并对校正集样品进行内部预测,其判别正确率为100%。更多还原

【Abstract】 Cow mastitis is one of the main factors that harm the dairy industry in the world.A rapid identification method of dairy cow mastitis based on two-dimensional near infraret correlation spectrum and pattern recognition was proposed.Firstly,milk samples of 40 Holstein cows were collected and analyzed in the range of 350～2 500 nm.The results showed that the normal 1-D NIR spectra of milk of healthy and sick dairy cows were similar,and the two-dimensional synchronous correlation spectrum analysis was carried out with the disease degree as the outer circle.A multi-dimensional partial least squares discriminant model(NPLS-DA) based on two-dimensional correlation spectrum characteristic matrix was established.The internal prediction of calibration set samples was carried out,and the discrimination accuracy was 100%.更多还原