[Simultaneous resolution of overlapped spectra of three kinds of organic compounds using a wavelet packet...

A wavelet packet transform-based generalized regression neural network (WPTGRNN) was developed to perform simultaneous spectrophotometric determination of p-nitroaniline, alpha-naphthylamine and benzidine. This method combines wavelet packet transform (WPT) with generalized regression neural network (GRNN) for improving the quality of noise removal and enhancing the ability of prediction. Wavelet packet representations of signals provided a local time-frequency description and separation ability between information and noise. The quality of noise removal can be further improved by using best-basis algorithm and thresholding operation. Generalized regression neural network (GRNN) was applied for overcoming the convergence problem encountered in back propagation training and facilitating nonlinear calculation. The GRNN is also advantageous in that the training process is much faster and without making any assumption about the form of the prediction model. By optimization, the wavelet function, decomposition level and smoothing factor of GRNN were selected. The partial least squares (PLS) method was used for comparative study. PLS method uses both the response and concentration information to enhance its ability of prediction. Three programs, PWPTGRNN, PGRNN and PPLS, were designed to perform relative calculations. Experimental results showed WPTGRNN method to be successful and better than others. Compared with GRNN method, the relative standard errors of all components between the actual and estimated values of mass concentration for WPTGRNN method decreased from 4.0% to 2.3%. Aniline type compounds are widely applied in industries such as chemistry, printing and pharmacy, and are one of the most important raw materials for synthetic medicine, dye, insecticides, polymer and explosives. Aniline-type compounds are highly poisonous, and can also cause cancer. Simultaneous determinations of aniline-type compounds are very important in environmental and industrial analysis.