Abstract
Metal-oxide-semiconductor gas sensors are widely used for gas concentration estimation and identification because of high sensitivity, good robustness and low price. To achieve good estimated performance, the feature extraction method is crucial for gas detection. However, the feature extracted from the transient part of gas response is required for fast gas concentrations estimation. In this paper, we proposed a new transient feature extraction method to reduce the estimated time of gas concentration. The linear regression was utilized to estimate gas concentrations. The result showed the estimation error of gas concentrations was less than 28.6%, and the average completion time was reduced up to 19 times. The proposed method showed superior performance to the conventional features.