硫脲是一种含氮量高、 毒性较大的潜在蛋白掺假化合物, 常规的实验室检测方法过程复杂、 效率低, 无法满足口岸对大批次散装奶粉样本质量快速筛查的需求。 为解决口岸抽样监管缺乏快速实时评价方法的难题, 利用自主搭建的便携式点扫描拉曼高光谱成像系统, 开发了一种简单高效的奶粉中硫脲现场可视化快速检测方法, 确保散装奶粉在进出口环节的精准监管。 研究以不同添加浓度(0.005%~2.000%)的硫脲奶粉混合物为样本, 分别用Whittaker平滑方法和自适应迭代重加权惩罚最小二乘算法(airPLS)消除光谱数据的背景随机噪音信号和荧光背景干扰, 峰值识别后对硫脲特征位移处的单波段数据进行二值化处理, 得到混合样本感兴趣区域的二值热图, 通过二值图中硫脲像素点的有无和坐标, 对奶粉中的硫脲进行定性判别和定位分析。 进一步分析得感兴趣区域内硫脲像素点数目与添加浓度的关系, 结果表明随着添加浓度的增加, 硫脲像素点数目呈线性增长趋势, 其中线性拟合的决定系数(R²)为0.991 3, 硫脲的最低检出浓度为0.05%。 在0.25%, 0.60%, 1.20%和1.50%的添加水平下, 利用像素点数目和线性拟合关系预测奶粉中硫脲的浓度, 结果显示预测浓度的相对误差范围为-9.41%~4.01%, 相对标准偏差小于7%。 该点扫描拉曼高光谱成像系统能在10 min内完成单个样本的检测, 结合软件控制系统, 实时对奶粉中的硫脲颗粒进行定性、 定量和污染分布分析, 方法简单高效、 准确性好、 灵敏度高、 稳定性强, 为口岸现场对散装奶粉中硫脲掺假物的实时快速检测提供了技术监管手段, 能显著提升口岸现场散装样本的监管环节质量评价的精准性, 为进口奶粉快速通关提供技术保障。

Thiourea is a potential protein adulteration compound with high nitrogen content and high toxicity. Conventional laboratory testing methods are complicated in-process and low inefficiency and cannot meet the port’s demand for rapid quality screening of large batches of bulk milk powder samples. In order to solve the problem of the lack of rapid real-time evaluation methods for port sampling supervision, this research uses a self-built portable point scanning Raman hyperspectral imaging system to develop a simple and efficient on-site visualized rapid detection method of thiourea in milk powder to ensure accurate supervision of bulk milk powder in the import and export process. In the study, thiourea milk powder mixtures with different additive concentrations (0.005%~2.000%) were used as samples. Whittaker smoothing method and adaptive iteratively reweighted penalized least squares (airPLS) were used to eliminate random background noise signal and fluorescent background interference of spectral data.. After peak identification, the single-band data at the characteristic displacement of thiourea is binarized to obtain a binary heat map of the region of interest of the mixed sample. The qualitative identification and positioning analysis of thiourea in milk powder can be carried out through the presence or absence and coordinates of the thiourea pixel in the binary map. Further analysis of the relationship between the number of thiourea pixels in the region of interest and the concentration of addition showed that with the increase of the concentration of addition, the number of thiourea pixels increased linearly, and the coefficient of determination (R²) of linear fitting was 0.991 3, the lowest detectable concentration of thiourea is 0.05%. Under the addition levels of 0.25%, 0.60%, 1.20%, and 1.50%, the number of pixels and the linear fitting relationship is used to predict the concentration of thiourea in milk powder. The results show that the relative error range of the predicted concentration is -9.41%~4.01%, the relative standard deviation is less than 7%. The point scanning Raman hyperspectral imaging system can complete the detection of a single sample within 10 minutes, combined with the software control system, real-time qualitative, quantitative and pollution distribution analysis of thiourea particles in milk powder. The method has the advantages of being simple and efficient, high sensitivity and stability, and good accuracy. It provides a technical supervision method for the real-time and rapid detection of adulterated thiourea in bulk milk powder at the port site and can significantly improve the quality evaluation of the supervision link of the bulk sample at the port, provide technical support for the rapid customs clearance of imported milk powder.