基于温度调制的气体传感器动态信号识别研究

摘要

Present a method for identifying three different kinds of industrial gases( CO,CH4 and C2 H5 OH),the sensor array consists of 4 SnO2 micro hotplate gas sensor doped Pt. Micro hotplate sensor heater adopts triangle wave,sine wave,saw tooth wave and pulse wave respectively. Then feature information of sensor dynamic response signal is extracted with short-time Fourier transform and combined with principal component analysis method, feature information is processed by dimensionality reduction,it is found that the influence of modulation period on gas species is smaller. Finally,three different gases are recognized by decision tree,recognition accuracy rate is high. Experimental result proves the correctness and feasibility of the algorithm.%利用4只掺杂Pt的SnO2微热板气体传感器组成传感器阵列来识别三种不同的工业气体（ CO， CH4和C2 H5 OH），传感器加热分别采用了三角波、正弦波、锯齿波和脉冲波四种温度调制模式。然后将得到的传感器动态响应信号进行短时傅里叶变换（ STFT）提取特征信息，并结合主成分分析（ PCA）方法将特征信息进行降维处理，发现调制周期对气体种类影响较小。最后利用决策树识别三种不同的气体，识别准确率较高。实验结果证明了该算法的正确性和可行性。