Hierarchical Temperature Imaging Using Pseudoinversed Convolutional Neural Network Aided TDLAS Tomography

Abstract

As an in situ combustion diagnostic tool, tunable diode laser absorption spectroscopy (TDLAS) tomography has been widely used for imaging 2-D temperature distributions in reactive flows. Compared with the computational tomographic algorithms, convolutional neural networks (CNNs) have been proved to be more robust and accurate for image reconstruction, particularly in the case of limited access of laser beams in the region of interest (RoI). In practice, the flame in the RoI that requires to be reconstructed with the good spatial resolution is commonly surrounded by low-temperature background. Although the background is not of high interest, spectroscopic absorption still exists due to heat dissipation and gas convection. Therefore, we propose a pseudoinversed CNN (PI-CNN) for hierarchical temperature imaging that: 1) uses efficiently the training and learning resources for temperature imaging in the RoI with good spatial resolution and 2) reconstructs the less spatially resolved background temperature by adequately addressing the integrity of the spectroscopic absorption model. In comparison with the traditional CNN, the newly introduced pseudoinversion of the RoI sensitivity matrix is more penetrating for revealing the inherent correlation between the projection data and the RoI to be reconstructed, thus prioritizing the temperature imaging in the RoI with high accuracy and high computational efficiency. In this article, the proposed algorithm was validated by both numerical simulation and lab-scale experiment, indicating good agreement between the phantoms and the high-fidelity reconstructions.