GA-PNN Based Multi-Gas Sensor Temperature Modulation Mode Study

The temperature modulation technique based on semiconductor gas sensors is a kind of multi-modal characteristic signal output with temperature dependence by periodically changing the heating temperature mode of the sensor, which can effectively address the low response rate and poor selectivity issues encountered in semiconductor gas sensors. To improve the detection characteristics of the sensor for four environmental test gases, CH4, CO, C 2 H 5 OH, and NO 2 , this paper uses rectangular wave temperature modulation mode for low-frequency dynamic performance test analysis, discrete wavelet transform and principal component analysis for dynamic information feature extraction and dimensionality reduction, combined with GA-PNN model for optimal identification. The results show that the selectivity of gas detection is improved in all the nine designed temperature modulation modes, and the response rate is improved significantly. Among them, the gas qualitative analysis accuracy can reach 100 ^% in the rectangular wave modulation mode with 4–5 V bias and a 20 s period, opening up a new semiconductor sensor detection path.