Study on selectivity of MOS-type gas sensor based on Prony's method and subspace identification method

在这项研究中，从半导体气体传感器的暂态响应中提取特征。对于作为一阶滞后系统阶跃响应的传感器响应，利用Prony方法可以将传感器输出近似为指数和。证明了该方法可以用子空间辨识方法来解释。模型参数可以估计为时间常数和增益，它们与气体的特性密切相关。这些参数可以用来构造用于气体识别的特征向量。用估计值作为特征，可以提高对气体的识别率。

In this study, feature extraction from a transient response of a semiconductor gas sensor is investigated. For the sensor response given as the step response of the first-order lag system, the sensor output can be approximated as a sum of exponentials by Prony's method. We have shown that the method can be interpreted by the subspace based identification method. The model parameters can be estimated as the time constant and the gain, which are strongly related to characteristics of gases. The parameters can be used to construct the feature vector for the discrimination of gases. Using the estimates as features, it has shown that the discrimination of gases can be improved.