简要分析由MEMS 工艺制成的新型微气体传感器阵列的原理及其优点,在此基础上,应用人工神经网络对气体传感器阵列的输出进行模式分类、识别,实现对单一或混合气体的有选择性探测,识别率可高达95 %。 关键词: MEMS 工艺; 微气体传感器阵列; 人工神经网络 Abstract : The principle and advantage of micro gas sensor array made by MEMS technology is analyzed sum2 marily , and based on it , arithmetic of BP neural network is applied to classify the electronic nose’s outputs and judge what the gases are. Not only single gas , but also mixed gases can be recognized selectively ,with a high dis2 cernment 95 %. Key words : MEMS process ; micro gas sensor array ; artificial neural network(ANN)