Laser absorption tomography of complex combustion fields based on finite element node strategy and adaptive edge optimization algorithm

Abstract

Laser absorption spectroscopy tomography is a technique for combustion diagnosis that provides two-dimensional temperature and species concentration measurements. However, its current implementation assumes a homogeneous distribution of discrete grids, introducing errors and uncertainties in the reconstruction results due to large gradients between adjacent grids and information loss within a grid. This work aims to develop a finite element node strategy (FES) to address these issues. To enhance the reconstruction performance, we propose an adaptive edge optimization algorithm (AEO) that can effectively reduce artifacts and preserve edges. We validate the effectiveness of the algorithm through comprehensive simulations and experiments. The simulation results show that the error of integrated projection value exhibits a trend of rise with increasing resolution, and it is also true that the more complex the reconstruction distribution is, the larger the error becomes. In the simulation, the maximum relative error by using the discrete grids strategy (DGS) reached 1.3 % for complex distribution, whilst it was only 0.1 % when the FES was used. Two representative flame phantoms were used in the simulation. The reconstruction error of 4.51 % was achieved in one phantom by using the FES over 5.11 % for using DGS, in the other phantom, the error was 3.14 % for the FES compared with 3.88 % for the DGS. With the addition of the adaptive edge optimization algorithm (FES-AEO), the temperature reconstruction errors were reduced to 1.49 % and 1.77 % respectively, and the characteristic peaks and edge values were closer to the true values. The experimental results show that when the resolution is 1.67 mm, the mean relative error of the DGS measurements is 1.26 %, while the FES-AEO is only 0.65 %.