【作者】 王岩；

【导师】 陈向东；

【作者基本信息】 西南交通大学 ， 信号与信息处理， 2006， 硕士

【摘要】 随着环境科学的发展及环境测量应用范围的不断扩大,成本低廉、性能良好的气体传感器已经成为测量领域的新方向。由于气体传感器自身的交叉敏感等物理缺陷,单—气体传感器无法对多气体进行准确的定性识别和定量检测。因此,将多个传感器组成阵列与模式识别方法相结合构造对气体分析的电子鼻系统,用该系统进行多种气体的识别和检测具有很大的实用价值,而模式识别技术对系统的行为起着关键性的作用。 本文主要对模式识别技术中的独立分量分析算法和神经网络算法进行了研究,介绍了独立分量分析算法的基础及独立性准则。引入了一种快速独立分量分析算法,将该算法应用到电子鼻系统的数据预处理中,可以将数据进行有效的预分类。将经过预处理后的数据样本作为BP神经网络的输入,进行仿真。预处理的数据是六个传感器组成的阵列对不同浓度的CO、CH₄、H₂三种气体进行测量所得。对不同浓度的CO、CH₄、H₂三种气体组成的三十个气体样本进行定性识别和对不同浓度的CH₄气体的三十个气体样本进行定量检测。 仿真结果表明,利用独立分量分析与BP神经网络相结合的模式识别技术,应用到电子鼻系统中是可行的,经过快速独立分量分析预处理后,减小了数据之间的相关性,而且BP网络结构得到简化,网络的收敛速度较快,而且测量精度较高。定性识别的准确率达100%,定量检测的准确率达97.19%。更多还原

【Abstract】 As the development of environmental science and the environmental detection technique, the cheap measuring gas sensors with good performance have become the new direction of measurement field. Caused by the physical shortcomings of gas sensors, it is impossible for a single gas sensor to identify multiple gases. So the electronic nose techniques based on gas sensor array and pattern recognition is becoming an important way in dealing with cross-sensitivity in gas analysis. The pattern recognition technology plays the crucial role to the behavior of electronic nose.The research mainly focuses on recognition technology of independent component analysis and neural network. It is important to introduce the foundation and independent criterion of the independent component analysis. FastICA（Fast independent component analysis） is brought and applied to data pre-processing in the electronic nose system. This pre-processed data is given a good classification for the gases. The data is inputted in the BP （back propagation） NN （neural network）, then do simulation .The original data is get from gas sensors array that is composed of six gas sensors. The 30 gas samples of different density CO、CH₄ and H₂ is carried on the qualitative identification and the 30 gas samples of different density CH4 is carried on the quantitative detection.The simulation results show that the pattern recognition technology which independent component analysis and neural network are unified is feasible in the electronic nose system. The data of per-processing with FastICA is eliminated the data correlation, and then the configuration of BP neural network obtained the simplification, the network convergence rate is quicker, moreover the recognition accuracy is higher. The qualitative identification rate reaches 100%, the quantitative detection rate reaches 97.19%.更多还原