基于近红外光谱法无损检测芦柑表面多种农药残留研究

以表面喷施不同浓度梯度的乙酰甲胺磷和毒死蜱混合农药的芦柑为研究对象。把两种农药以1:1比例混合,然后用自来水配置成1:100、1:300、1:500三种浓度农药溶液,分别喷施在192个芦柑表面上。采集近红外光谱数据后,应用NY/T761―2008方法 ,通过SP―6890气象色谱仪检测芦柑表面两种农药的残留量。对所得数据分别采用多元散射校正(MSC)和标准正态变量变换(SNV)两种光谱预处理方法 ,并结合联合区间偏最小二乘法(SiPLS)建立农药残留预测模型。结果表明:采用MSC的光谱预处理方法时建立的乙酰甲胺磷预测模型较优,其预测集相关系数R为0.8199;而采用SNV的光谱预处理时建立的SiPLS毒死蜱预测模型较优,其预测集相关系数为0.8434。可见,应用近红外光谱技术定量检测芦柑表面多种农药残留是可行的。

Study on Nondestructive Measurement Multiple Pesticide Residueson LuGan Mandarine Surface Based on Near-Infrared Spectroscopy

In the present paper,LuGan mandarine,sprayed with different concentration of the mixture of acephate and chlorpyrifos pesticide on surface were studied.The ratio of 1:1 mixture of two pesticides was diluted to1:100,1:300 and 1:500 solution with running water.The spectra of 192 LuGan samples were collected with a wavelength range of 350~1800nm.The pesticide residues on LuGan mandarine surface were determined by gas chromatography with NY/T761-2008 method.Two different spectral pretreatment methods of standard orthogonal variable transformation(SNV) and multiplicative scatter correction(MSC) were used for establishing the prediction models of synergy interval partial least square(SiPLS),which was determined the pesticide residue on LuGan mandarine surface.This study shows that the acephate model established by SiPLS after MSC pretreatment achieved the optimal performance.The correlation coefficient(R)of prediction set were 0.8199,and the chlorpyrifos model established by SiPLS after SNV pretreatment achieved the optimal performance.The correlation coefficient(R) of prediction set were 0.8434.The results shows that the method of using Near-Infrared spectroscopy to measurement multiple pesticide residues on LuGan mandarine surface is feasible.