姜桥,李宁,何来英,贾金柱,耿直,万劼,肖革新.压缩感知方法在食品安全风险监测中的应用[J].中国食品卫生杂志,2016,28(6):692-694. 本文二维码信息
二维码(扫一下试试看!)
压缩感知方法在食品安全风险监测中的应用
Application of compressed sensing method in food safety risk monitoring
投稿时间:2016-11-23  
DOI:
中文关键词:  压缩感知  混样  食品安全  检测  稀疏性  R统计  重构
Key Words:Compressed sensing  mixing samples  food safety  test  sparseness  R statistics  refactoring
基金项目:国家科技支撑计划课题:基于电子溯源的食品安全风险评估关键技术研究与应用(2015BAK3604);国家卫生计生委食品司委托课题:食品安全风险监测结果分析研究报告
作者单位E-mail
姜桥 国家食品安全风险评估中心,北京 100022;北京工业大学应用数理学院,北京 100124 qiaojiang_gogogo@163.com,xiaogexin@cfsa.net.cn 
李宁 国家食品安全风险评估中心,北京 100022  
何来英 国家食品安全风险评估中心,北京 100022  
贾金柱 北京大学,北京 100871  
耿直 北京大学,北京 100871  
万劼 首都经济贸易大学,北京 100070  
肖革新 国家食品安全风险评估中心,北京 100022 qiaojiang_gogogo@163.com,xiaogexin@cfsa.net.cn 
摘要点击次数: 112
全文下载次数: 54
中文摘要:
      压缩感知方法通过混样过程来减少样品的检测次数,从而提高检测效率,降低检测成本,缩短检测时间。方法 利用食品中污染物超标数据稀疏性的特点,将压缩感知方法应用于食品安全风险监测的样品检测。该理论的核心思想是通过混合待检测的样品,得到远少于原样品数的检测次数,然后根据相应重构算法由测量值重构原始数据。该算法可采用R统计软件实现。结果 用压缩感知方法重构125份原始样品的检测值,误差平方和为3.782 652×10-29,其中原始样品中117份低于检出限的样品全部精准重构,高于检出限的8份样品压缩感知重构值稍稍大于真实值,但误差极小,可以忽略不计。 结论 压缩感知方法可以通过混合样品来减少样品的检测次数,并可由少数检测值重构每一个原始样品的食品污染物含量。
Abstract:
      The compression sensing method reduces the times of sample inspection by sample mixing process, thus improving the sampling efficiency, reducing the sampling cost and shortening the sampling time. Methods Based on the characteristics of sparse data, the compression sensing method was applied to the food safety risk monitoring sample detection. The mixed samples were detected, and the original data was then reconstructed according to the corresponding algorithm with R statistical software. Results The detection values of 125 original samples were reconstructed using the compression sensing method, and the sum of squared errors was 3.782 652 × 10-29. There were 117 original samples lower than the detection limit were accurately reconstructed. The 8 samples which were higher than the detection limit were slightly larger than the real value, but the error is very small and can be neglected. Conclusion Compression sensing methods could reduce the testing number of samples by mixing samples and reconstruct the contaminant content of each original sample from several mixed samples.
查看全文  查看/发表评论  下载PDF阅读器