摘要
首先用碘酊染色显微镜体细胞计数法确定了100头不同患病程度的奶牛奶样的体细胞数,然后采用叉指电极在0.01～100 kHz频率范围对奶样进行了交流阻抗测试。在提出该测试系统等效电路的基础上用Zview软件获得了奶样电阻RS、电双层电容Cdl-T、Cdl-P等电参数,并分析了不同患病程度奶牛奶样的电参数特点。最后以镜检体细胞计数结果为标准,用电参数作为输入,建立了牛奶体细胞数的支持向量回归(SVR)预测模型,研究了体细胞数定量检测的可行性。结果表明,随着体细胞数的增加(即患病程度的增加),参数RS减小,而参数Cdl-T、Cdl-P呈非线性变化趋势。该定量预测模型对除N级以外所有奶样的体细胞数都有较高的预测精度,平均相对误差为29.40%。1级(隐性)、2级(较严重)和3级(严重)乳腺炎的检出正确率均达到100%。
A method based on electrical parameters and support vector regression for quantitative detection of somatic cell count （SCC） was exploded. Firstly, the SCC of 100 raw milk samples was confirmed by the standard microscope counting with iodine staining. And then, an interdigitated microelectrode was used in impedance measurements of milk samples in the frequency range from 0.01 Hz to 100 kHz. An equivalent electrical circuit of the milk samples was deduced for the measured impedance data. By using of Zview software, the electrical parameters of the equivalent electrical circuit were extracted and the relationships between these electrical parameters and the degree of mastitis were analyzed. At last, based on these electrical parameters and support vector regression （SVR） , a quantitative detection model of SCC was established and its feasibility was studied. It was showed that one of the electrical parameters of Rs was decreased with increasing of SCC, while other electrical parameters such as Cdl_T and Cd~_p presented a nonlinear trend. The SVR model was good for prediction of SCC with high precision except the negative samples. The average relative error was 29.40% , and the bovine mastitis detection rate for 1 （sub-clinical）, 2 （relative serious infection）, and 3 （serious infection） level of samples were 100%.
出处
《农业机械学报》
EI
CAS
CSCD
北大核心
2012年第8期164-169,共6页
Transactions of the Chinese Society for Agricultural Machinery
基金
国家自然科学基金资助项目(30871445)
关键词
奶牛
牛奶体细胞数
电参数
支持向量回归
定量检测
乳腺炎
Dairy cows, Milk somatic cell count, Electrical parameters, Support vector regression, Quantitative detection, Mastitis
