Three-dimensional temperature field tomographic reconstruction of diffusion flame based on information entropy termination criterion

Temperature represents a critical parameter in combustion system characterization, where radiation thermometry has become the predominant non-invasive technique for flame temperature monitoring. In the reconstruction of temperature fields, the inherent semi-convergence phenomenon of inversion algorithms requires particular attention to termination criteria, as these parameters crucially affect both computational efficiency and reconstruction accuracy. This study introduces an innovative termination criterion based on information entropy to address this challenge, enabling rapid identification of optimal inversion results while ensuring timely iteration cessation. Unlike conventional criteria relying on residual errors or predetermined iteration counts, the entropy-based approach exhibits broader applicability with reduced dependence on empirical parameters. Numerical simulations initially confirm the existence of semi-convergence during iterative reconstruction processes. Subsequent comparative analyses of various termination criteria under different measurement error conditions reveal the superior performance of the information entropy method. Robustness evaluations across varying signal-to-noise ratios and relaxation factors further substantiate the method's stability. Experimental validation was conducted using a multi-camera imaging system to measure three-dimensional temperature distributions in non-axisymmetric ethylene diffusion flames, achieving a 0.8925 correlation coefficient in back-projection verification that demonstrates practical effectiveness. Both numerical and experimental results confirm that the information entropy termination criterion provides reliable performance under measurement uncertainty conditions, offering significant advantages for simulation studies and practical combustion diagnostics.