基于盲信号分离的混合气体分析

Analysis of Gas Mixtures Based on Blind Signal Separation

结合盲信号分离算法和热催化传感器特性,对混合气体进行分析,讨论了混合气体分析的盲可分离性,并使用基于峭度的盲信号分离算法对CH4和CO的混合气体进行分析和验证。实验结果表明,此方法无须预先假设源信号的概率密度函数(PDF),可在学习过程中直接判断信号的PDF并进行分离,提高了混合气体分析的可行性,实验结果为CH4的测量误差为4.27%,CO的测量误差为3.51%。

Combing blind source separation algorithm and the character of heat catalytic sensor, the gas mixtures was analysed, the blind separability of the gas mixtures was discussed. The gas mixtures of CH4 and CO was made analyse and validate with the algorithm. Based on the kurtosis blind signal separation. The experimental results show that this method need not to firstly assume the probability density function （PDF） of source signals. It can estimate the PDF of the signal directly and carry on the separation in the study course. It enhances the feasibility of gas mixtures analysis. The result shows the survey error of CH4 is 4. 27%o and the survey error of CO is 3.51%.