Detection of chemical vapors using neural networks

This paper presents an application of neural networks as detectors in a class of chemical vapor sensors. One approach to chemical vapor detection is based on the steady state frequency shift of piezoelectric crystals in an oscillator circuit. The frequency of oscillation is a function of the surface wave velocity, which in turn, is affected by the changes in mass loading on the surface of the crystal. When the sensor is exposed to chemical vapors, the frequency of the oscillator changes with time. This paper discusses the detection of chemical vapors based on the dynamic behavior of the frequency shift. Neural network techniques have been used to identify the dynamic responses of the sensors and detect the chemical vapors generating the specific responses