Tunable diode laser absorption spectroscopy (TDLAS) tomography is a well-established combustion diagnostic technique for imaging two-dimensional cross-sectional distributions of critical flow-field parameters. As two key metrics in TDLAS tomography, reconstruction accuracy and efficiency are generally traded off to satisfy either the requirement of highfidelity image retrieval or rapid tomographic data inversion. In this paper, a novel quality-hierarchical temperature imaging network for TDLAS tomography is developed based on stacked Long Short Term Memory (LSTM). From limited line-of-sight TDLAS measurements, this network outputs two reconstructed temperature images, i.e. a coarse-quality image and a fine-quality image, with different numbers of network layers and consequently different computational costs. The coarse-quality image provides more timely temperature reconstruction, which can satisfy realtime dynamic monitoring of turbulence-chemistry interactions with a temporal resolution of tens of kilo frames per second. On the other hand, the fine-quality image, that can be stored and utilized for offline analysis and diagnosis, further details the temperature reconstruction with more accurate features. Both numerical stimulation and lab-scale experiment validated the accuracy-efficiency trade-off achieved by the proposed qualityhierarchical temperature imaging network.

Tunable diode laser absorption spectroscopy (TDLAS) tomography is a well-established combustion diagnostic technique for imaging two-dimensional cross-sectional distributions of critical flow-field parameters. As two key metrics in TDLAS tomography, reconstruction accuracy and efficiency are generally traded off to satisfy either the requirement of highfidelity image retrieval or rapid tomographic data inversion. In this paper, a novel quality-hierarchical temperature imaging network for TDLAS tomography is developed based on stacked Long Short Term Memory (LSTM). From limited line-of-sight TDLAS measurements, this network outputs two reconstructed temperature images, i.e. a coarse-quality image and a fine-quality image, with different numbers of network layers and consequently different computational costs. The coarse-quality image provides more timely temperature reconstruction, which can satisfy realtime dynamic monitoring of turbulence-chemistry interactions with a temporal resolution of tens of kilo frames per second. On the other hand, the fine-quality image, that can be stored and utilized for offline analysis and diagnosis, further details the temperature reconstruction with more accurate features. Both numerical stimulation and lab-scale experiment validated the accuracy-efficiency trade-off achieved by the proposed qualityhierarchical temperature imaging network.

Tunable diode laser absorption spectroscopy tomography (TDLAST) has been widely applied for imaging two-dimensional distributions of industrial flow-field parameters, e.g., temperature and species concentration. Two main interested imaging objectives in TDLAST are the local combustion and its radiation in the entire sensing region. State-of-the-art algorithms were developed to retrieve either of the two objectives. In this paper, we address the both by developing a novel multi-output imaging neural network, named as Spatially Progressive Neural Network (SpaProNet). This network consists of locally and globally prioritized reconstruction stages. The former enables hierarchical imaging of the finely resolved and highly accurate local combustion, but coarsely resolved background. The later retrieves a fine-resolved image for the entire sensing region, at the cost of slightly trading off the reconstruction accuracy in the combustion zone. Furthermore, the proposed network is driven by the hydrodynamics of the real reactive flows, in which the training dataset is obtained from large eddy simulation. The proposed SpaProNet is validated by both simulation and lab-scale experiment. In all test cases, the visual and quantitative metric comparisons show that the proposed SpaProNet outperforms the existing methods from the following two perspectives: a) the locally prioritized stage provides ever-better accuracy in the combustion zone; b) the globally prioritized stage shows turbulence-indicative accuracy in the entire sensing region for diagnosis of heat radiation from the flame and flame-air interactions.

No abstract