ANN modeling of micro-machined gas sensor signals

In this paper, an integrated micro-machined gas sensor array, associated with pattern recognition (PARC) techniques, such as artificial neural networks (ANNs), is designed. The proposed sensor design use a number of different sensitive films such as SnO₂, TiO₂, ZnO, or organic sensitive films to detect different gases. The application of micro-machined Si-based gas sensors in air quality management and emission control of internal combustion systems are very promising because of its compatibility. The reliability and accuracy of ANN predictions can be improved by systematic learning approach. The ANN models have the ability to describe the performance of complex and non-linear system behavior such as the non-linear signals produced by gas sensors. The use of ANN pattern recognition technique can lead to accurate modeling of individual gas concentrations in gas mixtures.