The Study Of The Algorithm Of Infrared Spectrometry Quantitative Analysis On Aqueous Solution Using Wavelet Transform

With the improvement of living standard and increase of social need, more requirements towards detection techniques are put forward hoping that detection techniques could be more convenient, more fast, more accurate, and non-polluting. Fourier transform infrared spectroscopy technology is one of the most useful chemical check technologies now. It has been widely applied in many production areas such as food hygiene, health, oil, chemicals because of numerous advantages, e.g. quick detecting, multi analysis, and no damage to the checked substances. And some detection techniques have been established as the standards of industry. However, most detected materials are based on solid materials. Analysis for liquid material, especially for mixed liquid material, still remains one of the difficulties in FTIR, and one of the hot spots of research.This paper is one of the sub-projects of the research project of quantitative analysis of mixed solution, and mainly discusses the problems of mixed solution of quantitative analysis, based on the ethanol-water solution. This paper first introduces basic theories of quantitative analysis using FTIR Lambert-Beer Law. The absorbency of Mixed solution is the addition of all components' absorbency in an ideal situation. However, for the mixed solution there are some objective factors which will have a negtive effect on the result of Lambert-Beer Law. This paper analyses four reasons about that and discusses two of the reasons which are chemical error and noise influence. Meanwhile, the absorbency spectra of water has a special feature, namely the spectra of water has high and wide peak near the 3400cm-1 wave number, and this peak can mask other peaks. This will lead to error of the quantitative analysis. And, components of mixed water solution will have a hydration with each other which cause components' absorbency spectra to have a deflection compared with the previous one. Due to all of these factors,common analysis methods in flourier transform infrared spectroscopy technology have a negative effect on water solution.Wavelet transform which is developed quickly recently is a new component of mathematics. Wavelet transform can analyze the signal in both frequency domain and time domain, and also can improve the resolution of the signal effectively. In this paper, on the basis of current analysis methods in fourier transform infrared spectroscopy technology, two algorithms of the quantitative analysis are discussed to improve the accuracy and robustness of ethanol-water solution. One of them is the improved analysis algorithms using advantages of the continuous wavelet transform which are linear and multi-resolution. Linear makes Lambert-Beer Law still stand after CWT. Furthernore, the high and wide peak of water will become lower than other peaks after CWT, and this effect reduces interference of the high and wide peak. On the other hand, multi-resolution also can reduce interference to quantitative analysis. Compared to the experimental,results it is found that the improved algorithm can improve the quantitative analysis' accuracy and robustness. Another one is that the infrared spectra is denoised using wavelet transform first, then partial least square method is used to analyze the ethanol-water solution. By the result of experiments, it is found that denoising infrared spectra using wavelet transform can improve quantitative analysis' accuracy and robustness as well.