Improving Selectivity of Gas Sensor Based on Rapid Temperature Modulation
To address the poor selectivity and cross-sensitivity of metal-oxide-semiconductor sensors,a fast temperature modulation method was used with an individual MEMS gas sensor to construct a virtual sensor array,which has lower power consumption and cost.First,the response signals to different gases were obtained under pulse temperature modulation and the modulation parameters were optimized.Then,a Support Vector Machine was employed to identify the types of different gases,and Support Vector Regression,Random Forest Regression,and Back-propagation neural network algorithms were employed to estimate the concentration of each gas.The results show that all four gases,H2,H2S,NH3 and C2H5OH,were correctly classified with concentration prediction errors of 19.5×10-6,3.7×10-6,0.2×10-6 and 19×10-6,respectively.This method improves the selectivity of individual gas sensors while reducing power consumption,providing ideas and solutions for on-site detection such as environmental monitoring and industrial production.
gas sensorselectivitytemperature modulationpattern recognition
国家科技期刊平台
NSTL
万方数据
