多光谱测温技术在燃气轮机的涡轮叶片、燃烧室等高温部件表面温度测量领域有着广泛的应用。然而，缺乏被测目标先验发射率信息，多光谱测温技术常产生较大误差，限制该技术的实际应用。为此，本文提出了一种融合发射率预识别的燃气轮机高温部件表面多光谱温度反演方法，该方法将欠定的辐射方程求解问题转化为目标函数的约束优化问题。首先基于灰体假设耦合广义逆矩阵法，获得发射率的取值范围，将其作为约束条件加入目标函数。进一步利用白鲨优化算法，实现发射率约束后目标函数的高精度温度求解。对不同温度下，六种不同的发射率模型的高温壁面开展了数值模拟，讨论了所提出算法参数的取值，并研究了其温度重建性能。最后，开展了高温壁面温度测量实验研究。数值模拟结果表明，800~1000 K温度下，温度反演最大误差为1.26%。在实验室条件下，测温结果与热电偶的最大偏差为1.96%，验证了融合发射率预识别和白鲨优化算法的表面温度场多光谱测量方法具有较高的精度和稳定性。

Multispectral thermometer is widely used in the field of temperature measurement of high temperature components of gas turbines such as turbine blades and combustion chambers. However, without the prior emissivity information of the measured target, the multi-spectral thermometer often has large measurement error, which limits its practical application. In this paper, a multi-spectrum radiation-based temperature inversion method combined with emissivity pre-recognition is proposed, which transforms solving the underdetermined radiation equations into a constrained optimization problem of the objective function. Based on the grey body hypothesis and generalized inverse matrix method, the range of emissivity is firstly obtained and added to the objective function as a constraint. The White shark optimization algorithm is further used to achieve the high-precision temperature solution of the constrained objective function. The key parameters and reconstruction performance of the proposed algorithm are discussed on the six simulated emissivity models at different temperatures. The simulation results showed that the maximum error of the temperature inversion is 1.26% at 800~1000 K. Finally, the temperature measurement experiments are carried out on a high temperature surface in the lab. The maximum deviation of the measured temperature between the proposed method and the thermocouple is 1.96%, which verifies that the proposed multi-spectral measurement method has high accuracy and stability.