【作者】 秦树基； 黄林；

【Author】 QIN Shu-ji. HUANG Lin(Dept. of Physics, Tongji Univ. , Shanghai 200092, China)

【机构】 同济大学物理系；

【摘要】 开发了一个能识别酒类的电子嗅觉系统.该系统基于一个由3只具有不同宽谱气敏响应特性的金属氧化物传感器组成的阵列,并由3层前馈神经网络用于分析传感器阵列的输出数据.神经网络采用Levenberg-Marquardt算法训练后,电子嗅觉系统能成功地识别酒精、烈性酒、葡萄酒和啤酒,正确识别率可达95%.更多还原

【Abstract】 A electronic olfactory system has been developed that is capable of discriminating among ethanol. spirit, wine and beer. The system is based on an array of 3 discrete in oxide semiconductor sensing elements, each possessing different characteristics whilst being sensitive to a broad spectrum of gases. Three-layer forward-feed neural network is used to analyzing output data from sensor array. and is trained by Levenberg-Marquardt algorithm. After training, ethanol, spirit, wine and beer identifications were successfully performed by the system, and the recognition probability reached 95%.更多还原