Application of principal component analysis for the characterisation of a piezoelectric sensors array

A piezoelectric sensor array was developed using four quartz crystals. For sensing materials gas chromatographic stationary phases were used. Factor analysis was performed and the weight of the factors was applied to characterise the selectivity of the sensing materials. The identification of some organic vapours was carried out by the sensor array. Principal component analysis was used and the signals were transformed into feature space. It was found that the transformed values of chloroform, acetone and cyclohexane were well separated. The data for the aromatic hydrocarbons were situated near the pixels of methanol and pentane. Organic materials can be identified in a mixture if the calculated factors are different. Interference has been found between benzene and toluene due to their similar factors. The selectivity of the chemical sensor array can be greatly improved using more sensing elements in one array and sensing compounds with very different chemical features. (C) 1998 Elsevier Science B.V.