The Development Of Combustible Gas Alarm Apparatus Using RBF Neural Networks

With the development of energy technologies, combustible gas are usedmore and more widely.It exsists not only the factories and large shopping mallsbut also the residential buildings. If the combustible gas is leakaged, it will causeimmeasurable loss of people’s life and propertoy. Therefore there must be a de-tection device which can detecte and judge the gas concentration to avoid the ac-cident.The paper firstly describes the researching background of the gas alarms aswell as the development status. Meanwhile compare the various gas sensors andanalyze the advantages and disadvantages. Based on the effects of temperatureabout the gas concentration and the RBF neural network learning,it proposes ascheme based on RBF neural network of combustible gas alarm controlling sys-tem. The system is mainly based on RBF neural network technology, compen-sates the temperature software of the sensor. And introduces the RBF neural net-work temperature compensation principle. It gives up the ordinary hardwarecompensation method so as to monitore the gas concentration more accurately.On the other hand, the article also make detailed instructions on the system’shardware and software design. And the system microprocessor used STC12sin-gle chip microcomputer, whose speed of calculation is12times that of the ordi-nary one. So it provides a powerful heart for the system. Pillar and digital tubedisplay driver chip adopt MCI4489(a professional digital tube display driverchip that is a Simple circuit and reliable performance) from Motorola. There is apiece of2.0inch TFT LCD screen on the main controller of the Interface display,so that it shows more intuitive and straight forward. At the same time, A switch-ing power supply circuit, sound and light alarm and linkage circuit, SD data stor-age circuit and RS485communication circuit, etc. are all included. It introduces the system of overall program flow diagram, Modbus protocol, data storage, etc.in the software part.Above all, in this paper, there is a detailed data analysize on the systemsoftware compensation. It makes a full statement on the error of the system, putsforwad some possible methods on improving the sensor precision and summa-rizes the system design.