Studies on blind separation of gas sensor array signals

Responses of a Micro-hotplate based integrated gas sensor array to CO and CH4 were measured with an automated gas sensor calibration system. Combining with the blind source separation(BSS techniques, the blind separability in gas mixture analysis was discussed. The widely used BSS approach-Independent Component Analysis(ICA) was adopted to verify the proposed method by analyzing the gas mixtures of CO and CH, The. analysis results demonstrate that BSS was an effective way to extract the information of gas components in mixtures, from which the gas concentrations can be estimated. The average relative quantification errors were 9.37% and 8.11% for CO and CH4, respectively, in the specified concentration ranges.