近红外测定稻谷水分定标模型验证研究

Research on calibration model verification for determination of moisture content in paddy with near-infrared method

通过对近红外测定稻谷水分定标模型进行验证,了解该定标模型用于快速检测稻谷水分的准确性、重复性是否满足日常监测的要求。分别选取有代表性的早籼稻和中晚籼稻样品各50余份,用标准方法和近红外方法进行测定,经对比两种方法测定结果的偏差,剔除其中异常值,然后进行校准标准差(SEP)计算;分别选取早籼稻验证样品集中水分高、中、低的3个验证样品,每种样测定10次,对测定结果进行重复性(sr)计算。结果表明,早籼稻验证样品集定标模型的系统偏差为0.15,校准标准差(SEP)为0.18%,3个验证样品的重复性(sr)分别为0.045%、0.066%、0.031%,中晚籼稻验证样品集定标模型的系统偏差为1.0,校准标准差(SEP)为1.1%。近红外分析仪所带定标模型能够满足大批量早籼稻水分快速检测的需求,对于中晚籼稻,定标模型需更新或重建,在满足标准要求后才能用于中晚籼稻水分快速检测。

Based on the calibration model to determine moisture content in paddy by using near-infrared spectroscopy,we verified the accuracy and repeatability for rapid detection of moisture in paddy to understand whether the method meet the requirementsof daily monitoring.Selecting representative samples of early indica rice and mid-late indica rice,each kind of more than 50 portion,we determined the moisture content with standard method and near-infrared spectroscopy,and compared the deviation of detection results,and carried out SEP calculation by removing the outliers;selecting three early indica rice check samples with high,middle and low moisture,respectively,we determined the moisture content for 10 times and carried out repeatability（sr）calculation.The results showed that：the calibration model＇s system deviation of early indica rice check samples set was 0.15,standard error of prediction corrected for bias（SEP）was 0.18%,three samples＇ repeatability（sr）were 0.045%,0.066% and 0.031%,respectively;the calibration model＇s system deviation of the mid-late indica rice check samples set was 1.0,standard error of prediction corrected for bias（SEP）was 1.1%.The calibration model of near-infrared spectroscopy could meet the requirement of rapid detection of moisture content in early indica rice,but for the mid-late indica rice,the model needed updating or rebuilding.