RESUMO
To recognize the key influential factors during the liquid preparation process of Tanreqing injection, the near infrared(NIR) spectra of the raw materials and the operating parameters of 24 batches of physical manufacturing were recorded as independent variables, and the total soluble solids contents and the light inspection acceptance rate of the final products were collected as dependent variables. The calibration models were developed using the partial least-square regression (PLSR) method, and the correlation coefficients between the independent variables and the dependent variables were calculated. For the quantitative models, the correlation coefficients for the calibration and inner cross validation of total soluble solids contents and the light inspection acceptance rate reached 0.911 9, 0.724 2 and 0.873 8, 0.795 9, respectively. Using the correlation coefficients diagrams, several key influence factors were preliminarily determined, and the physical significance were analyzed combined with production experience. This work demonstrated that NIR spectroscopy with PLSR algorithm could be used for the key influential factors recognization during the liquid preparation process of Tanreqing injection and can be popularized to solve similar problems..