首页|斜率截距校正算法在食用油酸值和过氧化值上的近红外光谱模型转移的研究

斜率截距校正算法在食用油酸值和过氧化值上的近红外光谱模型转移的研究

Near-Infrared Spectral Model Transfer of Acid Value and Peroxide Value of Edible Oil by Slope Intercept Correction Algorithm

扫码查看
良好的食用油近红外光谱分析模型以及不同仪器间的模型共享,能够提高模型利用率,可以满足食用油行业的发展需要.针对食用油酸值和过氧化值两个指标的模型转移问题,主要采用斜率截距校正算法,结合主仪器上建立的偏最小二乘法校正模型,针对从仪器上的食用油酸值和过氧化值的近红外光谱模型,进行了模型转移研究.研究表明,斜率截距校正算法在同为光纤探头扫描的实验仪器VERTEX 70和AntansⅡ(光纤探头部件)间有较好线性度与效果,其酸值和过氧化值的预测均方差从54.675 6和1 912.219 4,分别下降到了0.441 9和6.752.斜率截距校正算法在食用油酸值和过氧化值指标上对相同原理的仪器转移效果较好,并且食用油酸值的校正效果好于过氧化值校正效果.该研究结果对于食用油品质的快速分析模型的广泛应用具有重要意义.
Excellent near-infrared spectroscopy of edible oil quality and model sharing among different instruments can improve the utilization of the model and meet the development needs of the edible oil industry.Aiming at the problem of model transfer on the two indexes of edible oleic acid value and peroxide value,this paper mainly adopts the slop/intercept correction algorithm,combined with the partial least squares correction model on the main instrument,and studies the calibration transfer from the near-infrared spectroscopy model of edible oleic acid value and peroxide value.The results show that the slop/intercept correction algorithm has better linearity and effect between the VERTEX 70 and Antaris Ⅱ (fiber optic probe components)for the same optical fiber probe scanning.The root mean square error of prediction of the acid value and peroxide value decreased from 54.675 6 and 1 912.219 4,respectively,to 0.441 9 and 6.752.The slop/intercept algorithm has better effect on the same principle of instrumental oleic acid value and peroxide value,and the correction effect of edible oleic acid value is better than that of peroxide value correction effect.The results of this study are of great significance to the extensive application of the rapid analysis model of edible oil quality.

NIRslope/interceptcorrectionedible oilcalibration transfer

李天瑞、刘翠玲、位丽娜、吴静珠、孙晓荣

展开 >

北京工商大学计算机与信息工程学院

北京市食品安全与大数据重点实验室,北京 100048

近红外光谱 斜率/截距校正算法 食用油 模型转移

北京市自然科学基金北京市教委科研计划

4132008KZ201310011012

2018

中国粮油学报
中国粮油学会

中国粮油学报

CSTPCDCSCD北大核心
影响因子:1.056
ISSN:1003-0174
年,卷(期):2018.33(1)
  • 5
  • 3