注：本项目受国家自然科学基金资助 项目编号：60374049
作者：黄家锐 刘锦淮
单位：中国科学技术大学化学系 邮编：230031
中图分类号：TP212
文献标识码：A
文章编号：1006-883X(2005)01-0015-004
摘要：介绍了气体传感器动态检测结合神经网络识别空气中有机气体的新方法。这种方法利用单个SnO2气体传感器在方波温度调制的状态下实现了对多种有机气体的定性分析。在0.02Hz的调制频率、250℃～300℃的温度调制范围内，测得了传感器对不同浓度异丙醇、乙酰丙酮及其混合气体的动态响应值，再通过小波变换对单个周期测试信号进行特征提取，最后将提取的特征值输入神经网络进行网络训练和定性识别，识别的成功率高达100%。
关键词：动态检测；小波变换；有机气体；神经网络；定性识别
Qualitative Recognition Of Gases In Air Using Temperature-Modulated SnO2 Gas Sensor And Neural Network
Abstract: A new dynamic measurement method for the rapid identification and determination of volatile organic compounds in ambient air is described. For the qualitative recognition of the volatile organic compounds, one SnO2-based gas sensor operating in a rectangular temperature-modulation mode is required. The working temperature of the sensor is modulated between 250℃ and 300℃ and its dynamic respond to different concentrations of propane-2-ol, acetyl acetone, and propane-2-ol+acetyl acetone mixtures. The discrete wavelet transform (DWT) is used to extract important features from the sensor response. These features are then input to (neural) pattern recognition method. The success rate of species identification can reach 100%.
Keywords: dynamic detection; wavelet transform; volatile organic compounds; neural network; qualitative recognition
阅读全文
备注：2005年 第11卷 第1期