为了提高气体总发射率计算的准确性及效率，为高温燃烧中辐射换热的估算等工程应用提供可靠的发射率计算工具，针对CO2、H2O、CO 3种气体，基于伪洛伦兹线型重新生成了0.1~80 bar范围内更为准确的气体发射率查找表。基于新生成的气体发射率查找表的数据，训练了机器学习气体发射率预测模型，用于对气体发射率进行快速、准确地计算。结果表明，伪洛伦兹（pseudoLorentz）线型对高压气体光谱的计算明显优于Alberti等开发的谱线截断模型，能够更好地模拟CO2、H2O、CO 3种气体的谱线混合效应；基于伪洛伦兹线型开发的机器学习气体发射率计算模型，计算精度与逐线法直接积分算法相近，便于工程应用。

In order to improve the accuracy and efficiency of the calculation of the total gas emissivity, and to provide a reliable emissivity calculation tool for engineering applications, such as the estimation of radiative heat transfer in hightemperature combustion, a more accurate gas emissivity lookup table ranging from 0.1 to 80 bar was regenerated based on the pseudoLorentz spectral line profile for CO2, H2O, and CO. An accurate and compact machine learning prediction model was developed based on the regenerated emissivity data from lookup table for quick and accurate calculation of gas emissivity. Our investigations have shown that the pseudoLorentz spectral line profile can simulate the spectral line mixing effect of CO2, H2O, CO better, which can model the highpressure gas spectra significantly better than the Alberti cutoff model; the machine learning calculation model of gas emissivity developed based on the pseudoLorentz spectral line profile is as accurate as the direct linebyline (LBL) integration method and is suitable for engineering applications.

TK124

上海市自然科学基金（20ZR1426900）