Artificial intelligence techniques for peak homogeneity validation in liquid chromatography

A problem area in the interpretation of chromatographic data is the determination of the purity of a detected peak. The problem arises in deciding whether a peak is due solely to one component, or is caused by the simultaneous elution of two or more components. Whilst a number of multivariate statistical techniques are available to assess the number of components contributing to a given signal, they may require a significant amount of computer power to provide an answer, and are prone to noise-induced errors. As part of a programme of research aimed at developing an intelligent system for mobile phase optimisation, we have developed a knowledge based systems approach to the problem of peak homogeneity assessment. The expert system applies a number of simple numerical tests to the three-dimensional dataset produced by a photodiode-array detector and assess the results in light of the chromatographic signal. © 1989.