Transformed cluster analysis: an approach to the identification of gases/odours using an integrated gas-sensor array

Abstract

This paper deals with ‘transformed cluster analysis’, which can be used to analyse gas-sensor responses for the identification of individual gases/odours present in the ambient. The computational strategy for the analyses consists of transformation of data, graphical representation, formulation of the classification method, interpretation of results and quantification. The unknown gas/odour may be classified into five different probability groups according to the transformed responses obtained, i.e., real object, apparent object, probable object, doubt object and atypical object. This method also helps to evaluate the effect of dopants. The method is illustrated here with experimental data obtained in the laboratory. It is predicted that the technique can successfully be used for application-specific gas sensors.