Gas identification by wavelet transform-based fast feature extraction and support vector machine from temperature modulated semiconductor gas sensors

Semiconductor gas sensors are widely applied in agriculture and industrial fields for its low price and high sensitivity. For the physical shortcomings of gas sensors such as cross-sensitivity and lack of the stability, it is difficult to get steady and accurate result. In this paper we present a new strategy to extract features from the response of a thermally modulated semiconductor gas sensor, combined with Support Vector Machine (SVM) pattern recognition method for gas identification. A signal pre-processing method and wavelet decomposition transformation (DWT) were applied to extract features of a signal thermal modulated semiconductor gas sensor's response curves. Experiment result shows that the proposed method can perform well in discrimination of CO, H-2 their mixtures than traditional neural network.