In an odor sensing system, a flow-type measurement system has been used to
supply odorant vapors to the sensor cell since the measurement system is qu
ite simple and enables the rapid measurement. Although we have used the pea
k frequency shifts of QCM gas sensors during the vapor exposure as the sens
or outputs, the shapes of the transient responses may be useful to discrimi
nate odor samples. Moreover, the sensing system should be robust when the m
easurement condition such as supplied vapor concentration fluctuates. Then,
we proposed pulse vapor supply method and modelled the measurement system
to extract the information for odor identification. Since the variation of
the vapor concentration could be ignored in the pulse vapor supply method,
the system model became simple enough to be robust to the fluctuation. The
pattern separation among various samples with artificially fluctuated conce
ntrations was improved when the model parameters of transient responses aft
er the pulse vapor supply were used. (C) 2000 Elsevier Science B.V. All rig
hts reserved.