The Research Of Mixed-gas Test System Based On Wavelet Neural Network

Mixed-gas test system is mainly used in those fields which are related to our daily life, such as environment detection, casualty alarm, food sanitation, auto control and so on. So, it's very important.This paper designed and implemented a mixed-gas test system. By collecting the response data of the sensor array and analyzing it with WNN (Wavelet Neural Network), the system can identify the concentration of the mixed gas. The hardware of the system is made up of personal computer system, data acquisition system and gas sensor array. In addition, an application program based on MFC (Microsoft Foundation Class) was programmed. This application program provides an interface for users, so it's very convenient to control the data acquisition system and the users can feel the response of the gas sensor array in a straightway. Also, the user interface possesses the function of saving experimental data.Against the defects of traditional BP (Back-Propagation) algorithm, such as easily get into local minimum value and lower constringent speed, LM algorithm was introduced. Both algorithms were used in WNN training.Experimental results illustrate that the LM algorithm is superior to BP algorithm in training process and accuracy. Also, the identification effect is very good.