Study On The Temperature Modulation Technique Of Sensors In Air Monitoring E-Nose

Electronic nose(E-nose), rapidly developed since 1990 s, is an application of the artificial olfaction analog technology. By imitating the biological olfactory mechanism, it can analyze, recognize and detect complicated odours and most volatile gases. Generally,an E-nose system consists of sensor array,signal preprocessing module and pattern recognition unit. It has the characteristics of rapid detection and real-time online lossless or non-invasive monitoring. Taking the application of E-nose indoor air quality(IAQ) monitoring as the research background, to narrow down the scale of the sensor array of existing IAQ detector and thus lower the power dissipation and Cost of manufacture, this thesis focuses on the study of the application of temperature modulation in IAQ E-nose.Firstly, to obtain the data set under temperature modulation condition, a temperature modulation gas sensing system, which consists of voltage controlling part, sensor board and data acquisition part, is designed. The voltage controlling part is powered by DC power. The temperature modulation controlling is implemented through a N-channel MOSFET, which is working in linear region, controlled by the signal from a function generator. Then, experimental scheme is designed and conducted to obtain data.To solve the problems of classification of gases and prediction of concentration, the BP-ANN, SVM/SVR and Extreme Learning Machine(ELM) are used to establish models respectively. All the models are estimated in terms of classification accuracy or prediction relative error by cross-validation. For classification problem, all the models can obtain high classification accuracy. That means the application of temperature modulation technique can well capture the response features of gases. As to prediction problem, ELM is the best for HCHO and CO, obtaining lowest prediction error of 11.98% and 5.48%, and SVR obtains the lowest prediction error of 3.18% for NO2. The results of this work demonstrate that the temperature modulation gas system has a good performance in both classification and concentration prediction, and it is a feasible way to apply temperature modulation technique into IAQ detector so as to improve the availability of sensors, narrow down the sensor scale and lower power dissipation and cost of manufacture.