摘要
针对空气污染物氨气、乙醇、氨气乙醇混合气体,搭建在线检测电子鼻系统.采用不同的特征提取方法得出特征,并利用主成分分析(PCA)和线性判别式分析(LDA)做类别区分.结果显示,利用传感器响应最大值特征和LDA能更好地区分三类气体.利用最大响应值特征,采用多层感知器(MLP)神经网络和粒子群(POS)优化的支持向量机(SVM)对110个测试样本分类.结果显示,MLP神经网络的正确率为70%,POS优化的SVM正确率为96.364 0%.最后,根据Loadings分析,剔除了TGS2602,MQ138,MQ3传感器,优化了传感器阵列.结果表明,该在线电子鼻系统能够应用到这三类空气污染物分类.
Targeting air pollutants of ammonia, ethanol and the ammonia ethanol mixed gas, an online electronic nose system was established. Different feature extraction methods were used to obtain their features. Principal component analysis (PCA) and linear discriminant analysis (LDA) were used to classify them. The results showed that the three kinds of gases could be distinguished by using the maximum response characteristics of the sensor and LDA. Based on the maximum response value, a multilayer perceptron (MLP) neural network and an SVM optimized by particle swarm optimization (PO8) were used to test 110 samples classification. The results showed that the correct rate of the MLP neural network was 70% and the SVM optimized by POS was 96. 364 0%. Finally, according to the loadings analysis, the TGS2602, MQ138 and MQ3 sensors were removed and the sensor array was optimized. The online electronic nose system can be applied to the classification of these three types of air pollutants.
出处
《中国计量大学学报》
2017年第3期388-393,共6页
Journal of China University of Metrology
基金
国家重大科学仪器设备开发专项(No.2012YQ15008705)
浙江省科技计划项目(No.2015C33009)
关键词
电子鼻
特征提取
模式识别
传感器阵列优化
大气污染物
electronic nose
feature extraction
pattern recognition
sensor array optimization
air pollutant