摘要
采用近红外光谱仪,通过光学处理、数据处理和改进偏最小二乘法(MPLS)建立了快速测定高含磷量(321～632 mg/kg)和低含磷量(0～297 mg/kg)大豆油的近红外(NIR)模型。结果表明:高含磷量和低含磷量大豆油定标方程的交互定标决定系数(1-VR)分别为0.988和0.974,定标决定系数(R2)分别为0.992和0.980,定标标准误差(SEC)分别为2.420和2.512,交互定标标准误差(SECV)分别为2.538和2.678;现有数据预测标准偏差(SEP)分别为2.602和2.683;该近红外法在生产加工过程中可快速准确检测大豆油中含磷量。
The model of rapidly determining high phosphorus content（ 321 -632 mg/kg）and low phos- phorus content（0 -297 mg/kg）soybean oil was set up by modified partial least square regression（MPLS） after optics treatment and mathematical treatment using NIR. The results showed that the interactive cali- bration coefficients （ 1 - VR） for the soybean oil with high and low phosphorus content were O. 988 and O. 974, the average calibration coefficients of determination （ R2 ） were 0.992 and 0.980, the square standard errors of calibration（SEC） were 2.420 and 2.512, the square errors of interactive calibration（SECV） were 2. 538 and 2. 678 ,and the standard errors of prediction（SEP） were 2.602 and 2.683 ,respectively. The NIR method could be used to detect phosphorus content in soybean oil rapidly and accurately. Key words：MR; soybean oil; phosphorus content; modified partial least square regression（MPLS）
出处
《中国油脂》
CAS
CSCD
北大核心
2012年第7期82-84,共3页
China Oils and Fats
