摘要
研究了采用单只气敏元件通过变温加热和神经网络信息处理,对乙醇和醋酸混合气体的识别和浓度检测的方法。根据气敏元件的灵敏度随加热温度发生改变的特点,在气敏元件上施加锯齿波加热电压,测量了周期变温条件下25种不同浓度乙醇和醋酸混合气体的归一化电导—温度曲线。然后,采用由64个输入节点和2个输出节点组成的BP网络,以及Levenberg-Marquardt算法对测量数据进行处理。结果说明,该气体分析系统对乙醇和醋酸混合气体有较好的分类识别效果。
This paper researched the quantitative identification of individual gas concentrations （ethanol and acetic acid） in their gas mixtures using one gas sensor by periodic heating and neural network. The dependence of conduction of gas sensor on peri- odic heating temperature in 25 different concentrations of ethanol and acetic acid mixed gases was tested. In an artificial neural net- work ,64 inputs ,temperature changes of the sensors, and two outputs ,individual concentrations of the introduced gases were used. Levenberg-Marquardt training algorithm was performed as the training method of the neural network structure. The mixed gases tested curves were preprocessed, the results demonstrate that the proposed gas recognition systems are effective in identifying ethanol and acetic acid mixed gases.
出处
《仪表技术与传感器》
CSCD
北大核心
2010年第2期13-14,27,共3页
Instrument Technique and Sensor
