Weiliang Guo

Application of Principal Component Analysis-Artificial Neural Network in Near Infrared Spectroscopy for Determination of Compound Rifampicin Tablets

We have applied principal component analysis -artificial neural network (PCA-ANN) in near infrared (NIR) spectroscopy to synchronous and rapid determining the contents of rifampicin (RMP), isoniazide (INH) and pyrazinamide (PZA) in compound rifampicin tablets. Back-propagation (BP) networks which adopt Levenberg-Marquardt training algorithm have been developed. Via analyzing the NIR spectra matrix by principal component analysis (PCA) method, we have obtained the principal components (PC) scores. The original NIR spectra and PC scores were respectively used as input data. These developed BP networks have been optimized by selecting suitable topologic parameters and the best numbers of training. Compare with original NIR spectra, using the PC scores as input data, the capabilities of BP networks were much better. Using these optimized BP networks for predicting the contents of RMP, INH and PZA in prediction set, the root mean square error of prediction (RMSEP) are 0.00423, 0.00320 and 0.00608. These results are so satisfied and NIR spectroscopy technology is convenient, rapid, no pretreatment and no pollution that this method could be popularized in the in situ measurement and the on-line quality control for drug production.

Rapid determination of compound rifampicin tablets using near infrared spectroscopy with artificial neural network

This paper has investigated the application of near infrared (NIR) spectroscopy with artificial neural network (ANN) for synchronous and rapid determination of rifampicin, isoniazid and pyrazinamide in compound rifampicin tablets. We have developed Back-Propagation (BP) Networks which adopted Levenberg-Marquardt training algorithm and Log-sigmoid transfer function basing on NIR spectra of samples and contents of rifampicin, isoniazid and pyrazinamide. The degree of approximation, a new evaluation criterion of the network was employed, which proved the accuracy of the predicted results. The BP Networks have been optimized by selecting suitable topologic structure parameters and the best numbers of training. Using these BP Networks for predicting the amounts of rifampicin, isoniazid and pyrazinamide in prediction set, the root mean square error of prediction (RMSEP) are 0.00668, 0.00508 and 0.00680. These results demonstrate that this method is feasible. This method is convenient, rapid, has no pretreatment and no pollution.

Two modeling methods for near infrared spectroscopy nondestructive quantitative analysis of the protein contents in Coredyceps militaris mycelia powder

This paper presents a comparative study between Partial Least Squares (PLS) method and support vector regression (SVR) in modeling the relationship between the near infrared spectra (NIRS) and the protein contents in Cordyceps militaris mycelia powder samples. Both of the models were optimized by selecting the suitable spectra preprocessing methods and the best modeling parameters. And then the optimum models were obtained. The results demonstrated that the SVR model was superior to PLS model. The root mean square error of cross-validation (RMSECV), the coefficient relation between actual values and predictive values obtained by cross-validation (Rv) and root mean square error of prediction set (RMSEP) of the optimum SVR model were 0.0146, 0.9874 and 0.0130, which indicated that the stability, the fit and the predictive capability of the model were satisfied.