目的 利用近红外光谱（near-infrared spectroscopy，NIRS）结合化学计量学方法建立了麸炒白术Atractylodis Macrocephalae Rhizoma快速质量评价方法。方法 采用TQ Analyst软件采集麸炒白术不同炮制阶段饮片的NIRS，通过SIMCA软件，采用正交偏最小二乘-判别分析（orthogonal partial least squares-discriminant analysis，OPLS-DA）法建立不同炮制阶段白术饮片的定性判别分析模型。进一步以HPLC-DAD法测定的特征成分含量为参考值，以相关系数、校正均方根误差、预测均方根误差和性能指标为评价指标，筛选校正模型的最佳建模参数，并进行外部验证。结果 OPLS-DA结果显示，定性模型可预测4个不同炮制阶段的白术饮片。经筛选最佳模型参数后，发现当采用PLS法建立校正模型，检测8 033.95～4 030.50 cm⁻¹和9 677.05～9 194.93 cm⁻¹波段光谱，并采用“多元散射校正（multiplicative scatter correction，MSC）＋二阶导数（second derivative，2ndD）＋无平滑（no smoothing，NS）”对光谱进行预处理，主因子数均选择9时，校正模型最佳。且5个特征变量的R值均大于0.96，且外部验证误差均小于10%。白术特征成分含量随炮制时间的增加，而呈现先上升后下降的趋势。结论 所建立方法可对炮制过程中的麸炒白术饮片质量进行快速预测评价，为白术麸炒前后质量控制提供理论依据。

Objective To quickly assess the quality of Baizhu (Atractylodis Macrocephalae Rhizoma, AMR) processed by stir-frying with wheat bran using near-infrared spectroscopy (NIRS) in conjunction with chemometrics. Methods The NIRS of AMR processed by stir-frying with wheat bran at various stages of processing were obtained using TQ Analyst software. The qualitative discriminant analysis model of AMR processed by stir-frying with wheat bran in various processing stages using the orthogonal partial least squares-discriminant analysis (OPLS-DA) method was established using SIMCA software. Furthermore, the best modeling parameters of the calibration model were screened and externally verified using the correlation coefficient, root mean square error of calibration, root mean square error of prediction, and performance index as evaluation indexes, and the content of characteristic components was determined by the HPLC-DAD method as a reference value. Results The qualitative model was able to predict four distinct stages of AMR processed by stir-frying with wheat bran, according to the OPLS-DA results. The calibration model was the best when the OPLS method was used to establish the model and the number of principal factors was all selected as 9, with the detection spectra of 8 033.95—4 030.50 cm⁻¹ and 9 677.05—9 194.93 cm⁻¹ bands, using “multiplicative scatter correction (MSC) + second derivative (2ndD) + no smoothing (NS)” for spectra preprocess, according to the results of screening the best model parameters. Additionally, the R values of five characteristic variables were all greater than 0.96 and external verification errors were all less than 10%. The content of characteristic components of AMR showed a trend of increasing first and then decreasing with the increase of processing time. Conclusion This approach offers a theoretical foundation for quality control of AMR both before and after wheat bran frying, and it can rapidly anticipate and assess the quality of AMR processed by stir-frying with wheat bran during the processing process. The established method can rapidly predict and evaluate the quality of decoction pieces of AMR processed by stir-frying with wheat bran, providing a theoretical basis for the quality control before and after stir-frying with wheat bran of AMR.

R283.6

国家重点研发计划（2018YFC1707001）；杭州市农业与社会发展科研重点项目（202204A06）；宁波市“科技创新2025”重大专项（2020Z089）