Translation Modified Iteration And “Learning Mode” In Analyzing Overlapped Chromatography Peaks

The analysis of the separation of complex samples is one of the hotspots and difficulties at present in analytical chemistry. As the most commonly used tool, chromotography is usually combined with mass spectrometry or infrared instrument to analyze different sample system. Although the continuous development of the instrument, only relying on improving efficiency of the column and optimizing the separation conditions is difficult to achieve the complete separation. As the components are getting more and more complicated, the overlapping peaks are almost inevitable. Similar structure or property in the multicomponent system can often lead to spectra overlap and peak shape similarity. The need of more accurate quantitative results is one of the important and difficult study in chromatography.Traditional perpendicular method and tangent method have obvious defects, which encourages people to fine new means to resolve the overlapping peaks. With the development of chemometrics, it is more frequently applied in the field of resolution of overlapping peaks. Chemometrics methods include multivariate calibration method, Fourier transform, wavelet transform, artificial neural network, etc. Studies have shown that these algorithms can be effective resolutions, but there are complex calculation process, and will introduce too many subjective assumptions, which will lead to faults.This paper introduced a new approach for the quantitative resolution of overlapped chromatographic peaks. In this new approach, translation modified iteration(TMI) was proposed for overlapped chromatography. The capability of the approach to resolve peak date was evaluated for artificially overlapped chromatographic peaks and real unresolved pesticide samples within a wide range. Compare to other three popular models in resolution overlapped peaks thoroughly and systematically, TMI yields the minimum relative errors. The results showed that this method had a fast iterative speed and good convergence, performing good accuracy and reliability. With solid theoretical basis, it may be a promising method for resolving overlapped chromatography peaks.