节点文献

基于MR的主成分结合支持向量机鉴别蚕茧雌雄的研究

Study on the method of determinate pod Sex with PCA plus SVM by NIR

  • 推荐 CAJ下载
  • PDF下载
  • 不支持迅雷等下载工具,请取消加速工具后下载。

【作者】 颜辉陈斌金尚忠

【Author】 Yan Hui~(1,2) Chen Bin~1 Jin ShangZhong~3 1.JiangSuUniversity,JiangSu,ZhenJiang212013, 2.JiangSu University of Science and technology,JiangSu,ZhenJiang 212018, 3.ChinaJiLiangUniversity,ZheJiang,HangZhou,310018,

【机构】 江苏大学食品与生物工程学院中国计量学院信患工程学院

【摘要】 研究应用近红外光谱分析技术鉴别蚕茧雌雄的方法。获取蚕茧的近红外漫反射光谱, 其中校正集20枚,预测集10枚。对光谱进行软阈值小波消噪、5点一阶微分及正规化处理。提取主成分,计算累积贡献率,并分别对前2个及3个主成分作图。分别取1-10个主成分并结合不同的 e(10-1~10-9)用支持向量机建立预测模型。结果表明,小波消噪后,光谱平滑, 效果较好。从光谱的前3个主成分图中不能直接区分蚕茧的雌雄,且主成分相对分散,前 10个主成分累积贡献率才达85.23%。所建模型的预测正确率随着所取主成分个数增加而升高,随着 e 取值的下降而先升后降。当主成分数为6~10、e 为10-5时,预测正确率最高,达 94.7%。本研究为利用近红外光谱鉴别蚕茧雌雄的实践应用提供理论依据。

【Abstract】 Study on the method of determinating pod sex by NIR.20 samples from total 39 pods were select randomly for calibration;the other 19 pod was used as prediction.After diffuse reflection spectrum was get,it was denoised by wavelet,and was processed by 5 point 1 rank differential coefficient and normalization.The principal component of spectrum was picked up, accumulating contribution of principle component was acquired,and distribution of the 2 and 3 main principle component was draw.It was established that of the model of SVM with 1 to 10 principal components and 9 different e(10-1to 10-9).Results show that spectrum denosed by wavelet became smooth,the effect was good.It can not determinate the pod sex through distribution picture of 2 or 3 principle component,and it has many main principle component with low weight,the accumulating contribution of 10 main principle component was 85.23% only. The right rate of prediction was up with the number of principle component used,and it rise in the beginning and down afterward with decline of e.The best rate of prediction is 94.7% in term that e equaled to 10-5and 6 to 10 main principal components was adopted.This research will provide a novel way of determinating pod sex.

【关键词】 蚕茧雌雄 NIR主成分分析支持向量机
【Key words】 pod sexNIRPCASVM
  • 【会议录名称】 第九届全国计算(机)化学学术会议论文摘要集
  • 【会议名称】第九届全国计算(机)化学学术会议
  • 【会议时间】2007-08
  • 【会议地点】中国四川成都
  • 【分类号】Q96-3;S881
  • 【主办单位】中国化学会计算机化学专业委员会
节点文献中: 

本文链接的文献网络图示:

本文的引文网络