摘要
收集我国不同地区、不同品种、不同储藏时间的稻谷样品144份,应用近红外光谱(NIRS)技术研究了稻谷水分含量快速测定方法,在建立定标模型的过程中,探讨了光谱散射处理、数学(导数)处理等优化处理对定标模型的影响。结果表明:修正偏最小二乘法是建立稻谷水分含量测定定标模型的最适合数学方法,所建立的定标模型的相关系数(R)为0.9999,定标标准偏差(SECV)为0.04;55份样品外部检验的相关系数(r)为0.996,检验标准差(SEP)为0.072,标准方法与NIRS方法测定的水分含量之间的T检验值为1.685(P<0.05),两种方法测定结果无显著性差异,预测值与实测值的平均绝对偏差为0.03,说明所建立的稻谷水分含量测定的NIRS数学模型具有很高的预测准确性,可应用于稻谷品质分析的快速检测。
144 variety paddy samples collected from the major paddy growth areas in china were used to measure their moisture content by near-infrared spectroscopy（ NIRS）. The result showed that the calibration models developed by the partial least square（PLS） regression were the best. The correlation coefficient of calibration was 0.9999; standard deviation （SECV） was 0.04; regression squared （r） was 0.996; square error of pre- diction （SEP） was 0.072. T test value between the chemical standard methods and NIRS method is 1.685（P 〈 0.05）, which showed that there was no distinct statistic difference between the two methods. The averageabsolute deviation is 0.03. This proved that NIRS method is veracity to be used in practice to detect the moisture content in paddy.
出处
《粮油食品科技》
2008年第5期49-52,69,共5页
Science and Technology of Cereals,Oils and Foods
