基于近红外光谱与DPLS的潲水油快速鉴别方法

Study on Rapid Identification Methods of Waste Edible Oil Based on Near Infrared Spectroscopy and Discriminant Partial Least Squares

潲水油回流餐桌等食品安全问题越来越受到社会关注,探寻准确、快速、简便、高效且成本低廉的潲水油鉴别新方法成为食用油安全性能检测的新要求.对收集的82份潲水油和合格食用油样品进行了理化检测,鉴别出潲水油样品37份,合格油样品45份.从82份样品中随机选出24份样品作为第一组,余下58份样品作为第二组,以第一组24份样品作为校正集,建立判别偏最小二乘法(DPLS)模型,鉴别第二组58份样品,总体鉴别正确率为86.21%;再以第二组58份样品作为校正集,建立DPLS模型,鉴别第一组24份样品,总体鉴别正确率为95.83%.研究表明,基于近红外光谱与DPLS的潲水油快速鉴别方法可行,具有较好的鉴别效果.

A new identification method of waste edible oil based on near infrared spectroscopy was putted forward in this paper. Food safety problems, such as reselling waste edible oil as edible oil, were attracted more and more attention. In order to meet the identification requirements for edible oil safety, an accu- rate, fast, easy, efficient and low cost identification technology was needed. In this paper, 82 samples of waste edible oil and edible oil were collected, and 37 waste edible oil and 45 edible oil samples from the 82 samples were identified by chemical and physical methods. 24 samples randomly selected from 82 samples were first group samples, and the rest 58 samples was second group samples. In the first stage, the first group samples as the calibration set were used to build discriminant partial least squares（DPLS） model, which was used to identify the second group samples, and the accuracy rating was 86.21%. In the second stage, the second group samples as the calibration set were used to build discriminant partial least squares （DPLS） model, which was used to identify the first group samples, and the accuracy rating was 95.83%. The research showed that, the rapid identification method of waste edible oil based on near infrared spec- troscopy and discriminant partial least squares was feasible, its identification effect was preferably, and it can supply a new approach for the rapid identification of waste edible oil.