Study On Pure Spectrum Extraction And Retrieval Algorithm Of Gas Chromatography-mass Spectrometry

Gas chromatography - mass spectrometry (GC-MS) is an important tool of qualitative analysis of complex mixture samples. It combines advantages of gas chromatography and mass spectrometry when used alone and has greater advantage. It is widely used in environment, food, geological testing and other fields.The original GC-MS data is the basis for qualitative analysis, but it will contain the noise from the equipment, chromatographic column erosion substance, column matrix and so on. At the same time, if the sample contains some components whose retention time are close, their chromatogram peaks will be overlapped, and the mass spectrum of a component will contain some ion interference from other components. If the original GC-MS data is applied for qualitative analysis, the results will be unreliable. So, before the qualitative analysis, we need extract pure mass spectrum form original GC-MS data to improve the accuracy of qualitative analysis. Finally, we compare the spectrum obtained with the standard spectrum in the database to complete the qualitative analysis. The matching of unknown spectrum and standard spectrum is aided by computer, and the implementation of retrieval algorithm is the key.This paper aimed to improve and enhance the performance of spectrum extraction algorithm and retrieval algorithm to get accurate and reliable qualitative results. Two major works were completed in this paper. First, it was the research of spectrum extraction algorithm. In this paper, the GC-MS data was expressed as matrix. Then, spectrum extraction algorithm based on backfolding was proposed. This algorithm aimed to remove noise and extract independent components from overlapping peaks. Finally, the mass spectrums of pure components were obtained. It was the basis for qualitative analysis. Another work was the research of retrieval algorithm. A retrieval algorithm based on vector model was proposed. Firstly, unknown spectrum and standard spectrum were expressed as vectors. Then, the similarity between the unknown spectra and standard spectrum was obtained by comparing the similarity of their vector forms. Finally, the accurate and realizable results were achieved.By experiment, using the spectrum extraction algorithm and retrieval algorithm proposed in this paper, the analysis results of GC-MS data were consistent with the results of commercial software. The algorithms proposed in this paper had good performance, and quick and realizable qualitative analysis can be achieved.