中文摘要：

化学计量学算法为重叠GC-MS信号解析提供了有效手段.免疫算法（IAs）可根据标准样品的信号（色谱或质谱）对重叠GC-MS信号进行解析并得到重叠信号中各组分的信息.但是标样信号与实际测量信号之间的差别会造成解析结果出现偏差.针对标样信号与测量信号不一致的问题,作者提出了一种交替迭代算法用于重叠GC-MS测量信号的直接解析.该方法在不提供标样信号的情况下可以解析出组分的质谱和色谱信息.计算过程中,采用随机产生的质谱作为初始输入,利用最小二乘和IA算法交替计算质谱和色谱,直到满足终止条件.采用所建立的方法对40种组分的农药混合物进行了分析,使用快速升温程序在10 min保留时间内得到了全部组分的色谱和质谱信息.

英文摘要：

Chemometric methods have been proved to be a powerful tool for resolution of overlapping signals. Immune algorithm （IA） is one of the chemometric approaches for analyzing multi-component GC-MS signals. The method extracts the information of the components by iteratively eliminating standard information （chromatogram or mass spectrum） from overlapping GC-MS signals. In the primary IA, however, the standard signal of each component possibly contained in the mixture must be provided by measuring the standard or by theoretical simulation. When there is difference between the measured signals of the standards and the mixture, distortion and negative values will appear in the resolved chromatograms. In order to conquer the problem, an algorithm based on an alternative iteration of least squares fitting and IA was proposed in this work. In the method, the measurement of GC-MS was achieved with a very fast temperature program to make the ana- lytes to elute within a short retention time period, and then the chromatographic and mass spectral information of the compo- nents in the overlapping signal is calculated with the proposed algorithm. In the calculation, the algorithm takes random mass spectra of the components as the starting input, and then the information of each component is obtained with an alternating iterative process. In the iteration, the mass spectra are calculated by using the least squares fitting and the chromatographic profiles are resolved by IA. Furthermore, the non-negative and unimodality constraints are adopted in the calculation for im- proving the resolved results. The iteration stops when the remaining signal does not change. The feasibility of the method was validated by using a simulated GC-MS data matrix of a three-component mixture, and the practicability of the method was proved by resolving the GC-MS data of the 40-pesticide mixture. The results show that both the mass spectra and the chro- matographic information of the components were extracted from the overlapping signals, and