基于峭度的重叠峰解析新方法

Resolution of Overlapped Peaks Using a Novel Approach Component Analysis Based on Kurtosis

峭度是表征曲线陡峭程度的物理量。本实验提出了基于峭度的组分分析 (componentanalysisbasedonkurtosis,CABK)方法来解析重叠峰 ,从中分离出纯组分信息。这种新方法的优点在于半盲源分离 ,即只要判断出重叠峰所含组分的数目 ,就可从混合谱中分离出各组分的纯谱信息。将它用来解析模拟两组分重叠峰体系和酒样的GC MS混合谱 。

The separation of overlapped peaks is a popular and difficult problem in the resolution of spectra. Enlightened by the independent component analysis (ICA), the authors presented a novel approach named component analysis based on kurtosis (CABK) to identify the overlapped peaks. This method is semi-blind source separation, i.e. only if the number of components is determinated, the spectrum of pure component can be achieved. The identification effect is strongly decided by the kurtosis value. This method was applied to the simulation system and gas chromatography-mass spectra of wine, to achieve the satisfactory results.