高级检索+

基于近红外法的紫胶原胶主组分快速预测模型构建

Rapid Prediction Model of Primary Components of Sticklac Based on Near Infrared Method

  • 摘要: 采用化学法测定了紫胶原胶中的树脂、蜡质和色素的含量,用傅里叶变换近红外光谱技术采集原胶的近红外光谱数据,得到原始光谱,通过光谱预处理方法消除噪声,最后以偏最小二乘法(PLS)建立回归模型。最终得到紫胶中树脂、蜡质和色素含量的近红外光谱分析模型,其树脂、蜡质和色素的最优校正集决定系数(Rc2)分别为0.961、0.812和0.828,交叉验证标准误差(RMSECV)分别为1.96、0.28和0.065;验证集的最优决定系数(Rp2)分别为0.957、0.808和0.793,预测标准误差(RMSEP)分别为0.940、0.124和0.101。建立的紫胶原胶主组分近红外检测模型的准确度、稳定性及预测性能良好,为紫胶原胶主组分快速分析方法的建立提供了新思路。

     

    Abstract: Chemical methods were used to determine the content of resin, wax and pigment in the sticklac.Fourier transform near infrared spectroscopy (FT-NIR) was used to collect the near infrared spectra of sticklac, and the original spectra were obtained.Spectral pretreatment was employed to eliminate the noise, and partial least squares(PLS) was performed to establish the regression model.Finally, the analysis model of near infrared spectrum of resin, wax and pigment content in sticklac was established.The correction determination coefficients(Rc2) of resin, wax and pigment were 0.961, 0.812, 0.828, respectively. The root mean square errors of cross validation(RMSECV) were 1.96, 0.28, 0.065, respectively. The determination coefficients(Rp2) of validation sets were 0.957, 0.808, 0.793, respectively. And the root mean square errors of prediction(RMSEP) were 0.94, 0.124, 0.101, respectively.The results showed that the accuracy, stability and prediction performance of the model were good, which provided a new idea for the establishment of rapid analysis method for the primary components of sticklac.

     

/

返回文章
返回