Application of GA-RBF networks to the nondestructive determination of active component in pharmaceutical powder by NIR spectroscopy

A new method was applied to nondestructive quantitative analysis of pharmaceutical samples with different concentrations on the basis of the near-infrared spectral data. By the proposed method powerful radial basis function (RBF) networks can be produced based on a genetic algorithm (GA), which is applied to auto-configuring the structure of the networks to optimize the near-infrared wavelength regions used and variables employed in building radial basis function networks. Estimation and calibration of the sample concentration via NIR spectroscopy were made with the aid of genetic algorithm-radial basis function (GA-RBF) network models based on conventional spectra, standard normal variate (SNV), multiplicative scatter correction (MSC) and the first-derivative spectra, various optimum models of which were established and compared. Experiment results show that GA-RBF networks can give robust and satisfactory prediction and the GA-RBF model based on SNV preprocessing spectra was found to provide the best performance. Therefore, the proposed method may have significant potential for use in nondestructive quantitative analysis of pharmaceutical samples.