Temperature compensation study of infrared Sf6 gas sensor combining GA and wavelet neural network

Abstract

Ambient temperature changes have a nonlinear effect on infrared SF6 gas sensors, which leads to SF6 gas leakage and affects the normal operation of equipment. In this paper, based on the infrared differential detection technology and Lambert-Beer law for quantitative detection, a dual-channel pyroelectric detector is constructed for detection and analysis by using 3.75μm the measurement filter and 10.85μm the reference filter. In order to eliminate the nonlinear effect of the temperature change of the detection environment on the infrared gas sensor, a temperature compensation model based on the GA-WNN fusion algorithm is proposed to compensate the measurement error due to the temperature change of the detection environment by using its good nonlinear mapping as well as generalization ability. The experimental results showed that the maximum measurement error decreased from 333.93 ppm without temperature compensation to 80 ppm after temperature compensation in the range of detecting ambient temperature of 15-4°C and SF6 gas concentration of 0-2000 ppm. This method eliminates the need for additional external equipment to maintain the gas chamber temperature in dynamic equilibrium, thus avoiding an increase in the size of the gas sensor itself and the cost of fabrication. Compared to the traditional compensation method, this method does not require solving the fitting parameters sequentially and determining the temperature compensation coefficients segmentally. This reduces the amount of calculation and simplifies the process of temperature compensation, resulting in an excellent compensation effect.

Keywords

• gas sensor
• infrared differential detection
• lambert-beer law
• ga-wnn
• temperature compensation
• 68m12