基于TDLAS的痕量CO浓度检测系统及温压补偿

Detection system of trace CO based on TDLAS and temperature and pressure compensation

利用可调谐二极管激光吸收光谱技术（TDLAS）实现气体高灵敏度,高精度的非接触式检测。为避免二次谐波信号随环境中温度和压强的改变导致实测出现较大误差,需对测得的气体浓度进行温压补偿。实验以2332nm波长作为CO的中心吸收波长,以质量分数125×10^-6,1001×10^-6,1701×10^-6的CO作为实验气体。提出了利用BP神经网络补偿模型,并采用遗传算法（GA）与粒子群算法（PSO）优化BP,修正受温压影响的标气浓度,并进行了仿真测试对比。实验结果表明,采用PSO优化BP补偿效果最好,修正后的CO浓度平均相对误差约为1.55%,有效提高了CO气体检测系统的精度。

High sensitivity,high resolution and non-contact detection of trace gas can be achieve by tunable diode laser absorption spectroscopy（TDLAS）.But second harmonic signal will change with the temperature and pressure,resulting in a large error in the measurements.Therefore,it is necessary to perform the temperature and pressure correction for the measured values.The experiment uses 2332 nm wavelength as central absorption wavelength of CO and uses 125×10^-6,1001×10^-6,1701×10^-6 concentration of CO as the experimental gas.By changing temperature and pressure different concentration values of CO gas are got.In order to correct concentration and compensate the temperature and pressure,3 artificial neural network models are proposed and the simulation tests are compared.The experimental results show that the particle swarm optimization（PSO）algorithm has best compensation effect,and the average relative error of measured CO concentration is about 1.55%,so the gas detection accuracy can be largely improved.