摘要
本文用厚膜氧化锡半导体气敏传感器阵列对三种品牌卷烟烟气进行分析,详细阐述了实验过程及确定传感器阵列的组成方法,并从样本中用马氏距离选取合适的样本,用主成分分析法和神经网络聚类分析法对样本进行分析,主成分分析结果已较好的把各品牌的烟分开,神经网络对三种品牌烟的识别率分别为中华烟85%、云烟90%、红梅烟95%。
An thick-film tin oxide gas sensor array was used to analysis three different band cigarettes in this paper. The processes of the experiment and the component of sensor array were well expressed. The satisfying samples were obtained from initial samples by using Mahalanobis distance method. Principal component analysis (PCA) and neural network pattern recognition analysis were used to identify three cigarettes smoke samples. Good separation among the gases with different band samples was obtained using the principal component analysis. The recognition probability of the neural network was 85% for Zhonghua(中华),90% for Yunyan(云烟) and 95% for Hongmei(红梅).
出处
《化学传感器》
CAS
2000年第2期46-52,共7页
Chemical Sensors
基金
江苏省自然基金
关键词
传感器阵列
神经网络
卷烟
烟气分析
气敏传感器
gas sensor array, electronic nose system, neural network, cigarette smoke