作　　者： ; ;

机构地区： 华南理工大学轻工与食品学院制浆造纸工程国家重点实验室

出　　处： 《计算机与应用化学》 2005年第7期512-516,共5页

摘　　要： 盲信号分离算法可以实现重叠化学信号分离。在色谱分析中对盲信号分离命题进行解释,提出独立成分分析的独立成分对应分离的单位色谱峰,混合矩阵的元素对应各组分含量的单位比值。构造两组分的模拟色谱重叠峰作为计算验证,测定正已烷/正庚烷混合体系的色谱重叠峰作为实验验证。经过模拟和实验验证,对色谱分析中的重叠峰,进行各组分比例不同的多次测定,测定次数大于或等于组分数目,采用信号盲分离算法可以求出各组分的单位色谱峰,并确定重叠峰内各组分的相对含量。 Chemical overlapped signal may be separated in Blind Signal Separation （BSS）. In this paper, independent component analysis （ICA） , one of BSS methods, was used for separating independence chromatograph peaks from overlapped peak, those separated peaks conform to the quota relations, and keep immobile retention time. BSS proposition was explained in chromatograph analysis; it is proposed that independent components in ICA correspond to separated unit chromatograph peaks, elements in ICA hybrid matrix correspond to unit ratios of each matter components. Overlapped chromatograph peaks including two simulation components were con- structed to be computing confirmation; Hexane/Heptane system was determined to be an experiment confirmation. Confirmed by the simulation and experiment, after many measures which times is more than or equal to the component number and component quotas are different, the chromatograph overlapped peaks are separated to each unit peaks of components by BSS, each component comparative contents in overlapped peak are certain and relative error is less than 5%.

关 键 词： 盲信号分离 独立分量分析 重叠化学信号 色谱

领　　域： [理学] [理学]