热声耦合振荡燃烧问题是发动机燃烧室在研制过程中经常遇到的关键难题之一，其早期诊断和控制有助于降低研制成本并提高发动机的可靠安全性能。激光吸收光谱层析成像技术是测量燃烧参数分布的有效手段之一。但现有的激光吸收光谱层析成像重建算法由于光路直进的假设，无法满足振荡燃烧场强偏折效应下的高精度重建需求。针对上述问题，本项目提出基于非直线路径积分的激光吸收光谱层析成像重建方法。首先，研究基于自适应步长光线追迹法的偏折传输过程建模方法，并结合求得的偏折特征参数分析强偏折效应对层析成像的影响机理；其次，研究基于偏折路径反演的激光吸收光谱层析重建方法，结合数据重排及修正方法建立偏折路径下的光谱吸收率分布重建方程组，研究该病态重建方程组的稳定求解优化算法，并进行重建方法的仿真验证。最后，在振荡燃烧装置上对所提出的重建方法进行实验验证。本项目的研究内容可为振荡燃烧及其它强偏折燃烧场的精细化诊断提供技术支撑。

Thermo-acoustic coupled oscillation combustion is one of the key problems encountered in the development of engine combustion chambers. Its early diagnosis and control can help reduce development costs and improve the reliability and safety of the engine. Laser absorption spectroscopy tomography is one of the effective methods to measure the distribution of combustion parameters. However, the existing reconstruction algorithm of laser absorption spectrum tomography cannot meet the high-precision reconstruction requirements under the strong deflection effect of the oscillating combustion field due to the assumption that the optical path is straight. In response to the above problems, this project proposes a nonlinear-path-based reconstruction method. First, the deflection transmission process is modeled based on the adaptive-step ray tracing method, and the influence mechanism of the strong deflection effect on tomography is analyzed in combination with the obtained deflection parameters. Second, a laser absorption spectrum tomography reconstruction method based on inversion of deflection path is studied. The reconstruction equations of spectral absorption rate is established combining with the data rearrangement method. The stable solution optimization algorithm is researched for the pathological reconstruction equations. Finally, on the basis of simulation verification, the proposed reconstruction method is experimentally verified on an oscillating combustion device. The research content of this project can provide technical support for refined diagnosis of oscillating combustion and other strong deflection combustion fields.