摘要
研制了一套由51单片机和金属氧化物半导体气敏传感器阵列组成的便携式电子鼻系统,介绍了系统的工作原理和试验过程。对不同气体通过传感器阵列产生的响应信号进行了数据采集,并对各条响应曲线进行了特征分析,得出了适合单片机处理的特征向量。在低成本的51单片机中编制了神经网络识别程序,神经网络训练过程则在PC机上完成。在51单片机中用排序法和BP神经网络对样本进行分析识别,模拟人和动物的气味学习、记忆过程来学习气味特征、识别不同气味。文中阐述了详细的实验过程和部分气体特征向量,实际识别正确率为90%以上。
A portable electronic nose system made up of 8051 micro-controller and metal-oxide semiconductor (MOS) gas sensor array is developed and the operational principle and experiment process of the system are introduced. The response signals generated by different gas as they pass sensor arrays are collected and the characters of various response curves are analyzed; finally, the feature vectors suitable for the microcontroller to deal with are obtained. The neural network recognition program is burned into the low-cost 8051 micro-controller while the neural network training procedure is completed on a PC. In sample identifi- cation, ranking method and BP neural network are used in 8051 micro-controller to simulate the learning and memorizing processes of human beings and animals; accordingly, the features of smell are studied and different smells are identified. The detailed experiment process and characteristic value of part of gases are described in the article, with a real correct recoznition rate hieher than 90%.
出处
《电子器件》
CAS
2008年第5期1514-1517,共4页
Chinese Journal of Electron Devices
基金
江苏省高校自然科学基础研究项目06KJB510135
扬州大学科技创新基金资助
关键词
51单片机
电子鼻
排序法
神经网络
气体识别
8051 micro-controller
electronic nose
ranking method
neural network
gas recognition