An intelligent gas sensor system for the identification ofhazardous airborne compounds using an array of semiconductor gas sensorsand Kohonen feature map neural networks

摘要

An intelligent gas sensor system for identification andquantification of hazardous airborne compounds has been developed. Asgas detecting devices semiconductor gas sensors with partiallyoverlapping selectivity have been used. Because of the low selectivityof semiconductor gas sensors, one gas sensor alone can give no accuratestatement concerning type of gas and the actual concentration. Applyingan array of semiconductor gas sensors in combination with Kohonenfeature map (KFM) neural networks, unknown species of gases can beidentified and quantified if the gas sensor system has been calibratedwith this sort of gas earlier. The influence of different networkparameters, e.g. the number of nodes in the network or the number ofpattern vectors used to train the KFM have been studied. It has beenfound, that the KFM is able to identify all compounds which have beenused for calibrating the gas sensor array and for training the KFM