Neural networks and abductive networks for chemical sensor signals : a case comparison

Abstract Both artificial neural networks (ANNs) and the abductory induction mechanism (AIM) have been proven to be suitable for the evaluation of signals from chemical sensors with strong interactions of gas components on the catalytically active surface. The algorithms allow the calculation of calibration curves for a multisensor. AIM yields a good mean approximation within a very short time; ANN covers a broader concentration range with an adequate approximation. The signal evaluation of a set of two pellistors and another set of six MOSFETs is used for illustration.