Toward the realization of an intelligent gas sensing system utilizing a non-linear dynamic response

We report on a trial to construct an intelligent gas sensing system based on the information embedded in a non-linear dynamic response, an application that has possibilities for various kinds of practical usage. By applying a sinusoidal voltage to a heater attached to SnO2, a characteristic time-dependent trace of the sensor resistance is obtained as a response to environmental gases. In order to evaluate the characteristic response in a quantitative manner, Fast Fourier Transform (FFT) is performed for the dynamic response. Higher harmonics, obtained by performing the FFT, were processed using an Artificial Neural Network (ANN). It is shown that with these procedures one can simultaneously distinguish and quantify individual gas components. Actually, we show that eight different gases (methanol, ethanol, acetone, diethyl ether, benzene, iso-butane, ammonia and ethylene) as well as natural air can be identified with a single sensor, and can also be quantified with an accuracy of less than 30%. It has been confirmed that our system exhibits long-term reproducibility, and the ability for discrimination and quantification.