城市危险品气体检测对于城市安全至关重要。利用SnO2,In2O3,SnO2/NiO,SnO2/In2O3,SnO2/Pa,SnO2/Sb 6种旁热式气敏元件组成的气体传感器阵列,分别采用BP神经网络、支持向量机(SVM)和极限学习机(ELM)3种气体辨识方法实现对城市危险气体中常见的3种气体(NH3,HCOH,C7H8)进行定性的识别。测试结果表明:基于SVM与ELM的气体辨识技术对于含有低浓度甲醛的混合气体定性识别率达100%,且在收敛速度、泛化性能等方面较BP神经网络有明显提高。
Detection of urban hazardous gases are very important for urban safety.In this paper,the gas sensors array was composed of six types of side-heating gas-sensitive sensors which are made of SnO2,In2O3,SnO2/NiO,SnO2/In2O3,SnO2/Pa,SnO2/Sb.The three kinds of gas recognition mothods,Back-Propagation NN(BP),extreme learning machine(SVM) and extreme learning machine(ELM) were uesd to indentify three kinds of urban hazardous gases(NH3,HCOH,C7H8).The detection result shows that gas recognition mothods based on SVM and ELM are improved than BP in generalization performance and convergence speed,the qualitative recognition rate to low concentration of formaldehyde in the mixed gas is 100%.