基于GA-WNN温度补偿的红外CO2气体传感器系统研究

为了提高红外CO2气体传感器的探测灵敏度和精度，首先基于计算流体动力学（CFD）仿真计算，研究了传感器腔内气体辐射功率吸收效率与腔体结构之间的关系，模拟结果表明：当圆柱腔体的直径与内壁反射率固定时，腔体结构存在最佳腔长可使传感器红外辐射功率吸收效率达到最大。然后基于CFD仿真的结果设计和实现了CO2气体传感器，并开展了实验比对与验证，进而着重研究了环境温度对气体测量结果的影响。实验结果表明：在5~45oC温度范围内，传感器在0~2000 ppm浓度范围内的测量误差随着温度升高而显著增大。最后采用遗传小波神经网络算法（GA-WNN）对传感器进行了温度补偿，数据融合补偿后传感器的温度漂移得到了较好的抑制，其绝对误差小于±70 ppm，在非样本温度点下，整体平均误差小于±100 ppm，表明CO2气体传感器的测量精度得到了较好的提升。

Research on Infrared CO2 Gas Sensor System withTemperature Compensation Based on GA-WNN

In order to improve the detection sensitivity and accuracy of infrared CO2 gas sensors, the relationship between absorption efficiency of gas radiation power in the sensor cavity and the cavity structure was firstly studied based on computational fluid dynamics (CFD) simulation calculations. The simulation results show that when the diameter and inner wall reflectivity of the cylindrical cavity are fixed, the cavity structure has the best cavity length, which can make the absorption efficiency of infrared radiation power of the sensor reach the maximum value. Based on the CFD simulation results, the CO2 gas sensors were then designed and implemented. Experimental comparisons and verifications were conducted, and then the influence of ambient temperature on gas measurement results was investigated. The experimental results show that the measurement error of the sensor increases significantly with the increase of temperature in the range of 5 ~ 45 oC and in the concentration range of 0 ~ 2000 ppm. Finally, the temperature drift of the sensor was compensated by genetic wavelet neural network algorithm (GA-WNN). The temperature drift of the sensor after data fusion compensation was well suppressed. The absolute error was less than ±70 ppm. At the non-sample temperature point, the whole average error is less than ±100 ppm, which indicates that the measurement accuracy of the CO2 gas sensor has been improved.