基于气体传感器动力学模型的细菌分类研究

Research on classification of bacteria based on kinetics model for gas sensor

对于气体传感器获得的样本数据,常规的处理方法是基于样本的表象来提取特征进行分类,具有固有的局限性。基于热电子发射理论和等温吸附理论,对气体传感器的电压响应值与温度、样本浓度等参数建立数学方程,将方程简化,构造了简易的响应动力学模型。通过主成分分析,降低样本维数。将模型向降维后的样本数据拟合,可得到模型的一组系数,作为样本的特征值。将特征值集合运用模式识别方法进行训练,测试分类性能。实验结果显示:分类预测的准确率较高。

For sample data acquired by gas sensor,ordinary way is to extract features based on surface pattern of samples,so as to classify,so it has inherent limitation. Based on thermionic emission theory and adsorption theory,a mathematic equation can be built to demonstrate the connection of voltage response values of gas sensor with temperature,sample concentration,etc. A simple model of gas sensor＇s response kinetics can be derived by simplify the equation before. Through the method of principal component analysis（ PCA）,the dimensions of samples is reduced. A group of parameters,which will be treated as features of samples eventually,can be settled when fitting the model to the samples. A prediction model will be trained and benchmarked by using the feature sets. The result shows that the kinetics model has a satisfied prediction accuracy in classification,this is for the reason that the model is set up based on reflection of nature of sensor response.