Gas Sensor Array Pattern Recognition Algorithms And Hardware Implementation

With the development of science and technology, people now pay more and moreattention to their health. Single gas sensor cannot detect the mixed gas with thecross-sensitivity characteristics. Consequently, it is very inportant to design a devicethat can quickly and accurately to detect the toxic and harmful gases. The dissertation isbased on such background. In the dissertation, in order to guide the acutal algorithmused in the system, we introduce different kinds of algorithms, such as PrincipalComponent Analysis(PCA), Back-Propagation, Support Vector Machine, et al. Researchconsists of the following:(1) Realization of the pattern recognition algorithmConsider the different kinds of embedded operating systems, we finally choose theArm-Linux operating system. Using the platform of Qt-Creator and the third-party classlibraries, design the communication interface ande interactive software system.(2) Drawing of the hardware circuitComparing different core chips used in the embeded platform, we finally choosethe Samsung S3C6410as the core chip of the system. The system is divided into twoparts which are core board and expansion board. Core board contains the minimummodules of the system and power modules, memory modules. Expansion board includescommunications module, display module, alarm module.(3) Analysis of the dataThrough the gas sensors, analysis the data with the design of the software,and alsocompare the result of different sensor responses in order to improve the applicability ofthe algorithm.