油用牡丹单粒种子含油量NIRS模型的建立

为了建立油用牡丹单粒种子含油量的近红外测定模型,便于高含油量单株的选育,采用索氏抽提法测试了200份油用牡丹凤丹单粒种子的含油量,并应用近红外反射光谱技术(NIRS)采集了200份样品的光谱数据,通过偏最小二乘法(PLS)和主成分回归法(PCR)构建了油用牡丹单粒种子含油量的数学模型。结果表明,索氏抽提法中,均匀粉碎后的油用牡丹籽样品干燥烘焙条件为105℃ 2 h,牡丹籽抽提时间为20 h,测出的含油量变化范围在10%~28%之间,籽油含量基本符合正态分布。NIRS法构建的模型最佳参数为:采用PLS法,光程固定,一阶导数消除背景,数据平滑处理采用Norris derivative filter的方法,平滑参数选用5和3。内部交叉检验校正相关系数r₁为0.980 1、预测相关系数r₂为0.957 6、校正均方根误差(RMSEC)为0.463、预测均方根误差(RMSEP)为0.705。外部检验相关系数达 0.957 6, 平均误差小于3%。本试验所构建的牡丹单粒种子含油量的NIRS模型可靠,可以用于分析油用牡丹单粒种子的含油量。

Establishment of NIRS Model for Oil Content in Single Seed of Oil Peony

In order to establish the near infrared model of measuring oil content in single seed of oil peony, which is convenient for breeding with high oil content. Oil contents of 200 single seeds of Paeonia suffruticosa Fengdan were measured by Soxhlet extraction method. Spectral data of these 200 samples were collected by near infrared reflectance spectroscopy (NIRS). The mathematical model of single-seed oil content in Paeonia suffruticosa was established by the methods of partial least squares (PLS) and principal component regression (PCR). The results showed that: with Soxhlet extraction method, dry baking conditions of oil peony seeds were 105℃ for 2 h, and extraction time was 20 h. The oil contents ranged from 10% to 28%, which was basically conformed to normal distribution. The best parameters for NIRS model were as follows: PLS method, fixed optical path, first derivative to eliminate background, Norris derivative filter method for data smoothing, 5 and 3 for smoothing parameters. The correlation coefficient of internal cross validation test r1 was 0.980 1 and the prediction correlation coefficient r2 was 0.957 6. The root mean square error of calibration (RMSEC) and the root mean square error of prediction (RMSEP) were 0.463 and 0.705 respectively. The determination coefficient of validation was 0.957 6, and the average error was less than 3%. The results showed that the NIRS model established in this study was reliable and can be applied in single-seed oil content analysis of Paeonia suffruticosa.