Analyzing the Responses of a Thermally Modulated Gas Sensor Using a Linear System Identification Technique for Gas Diagnosis

A novel approach to the problem of diagnostic data extraction from the responses of a thermally modulated resistive gas sensor (RGS) is presented. The RGS affected by a target gas (TG) is considered a black box dynamic system. The input to the system is the time-varying voltage applied to the heating element of the RGS, and the transient response of the RGS is the output. The structure of the defined system varies with the nature and concentration of the prevailing TG, and the parametric system identification techniques employed reveal system parameters differentiated only by the existing dissimilarities between the TGs. The discriminative information content of these parameters is, then, extracted by standard mathematical tools and utilized for TG recognition. Air contaminated with four different combustible vapors, methanol, ethanol, 2-propanol, and 1-butanol, each at 13 different contamination levels, was used to define 52 different systems. In each case, the transient response of the system to a staircase voltage waveform input was recorded. Computer modeling, based on
autoregressive

moving

average

with

exogenous

input
(ARMAX) model, rendered different sets of system parameters which afforded feature extraction and TG classification by standard mapping tools. The method was verified by the successful classification of unknown TGs at undetermined contamination levels.