A novel parametric level set method coupled with Tikhonov regularization for tomographic laser absorption reconstruction

Abstract

• Clear topological boundary of the combustion reaction flow is obtained. • Variations of temperature and species concentration in the chamber are diagnosed. • The accuracy, robustness and noise immunity of proposed method are the best. • The diagnosis of reaction flow with a large boundary gradient is highly accurate. Temperature and concentration distributions are essential parameters for understanding physicochemical processes in the in-cylinder/in-chamber engine and developing efficient combustion technology. However, due to the limited measurement space in the combustion chamber and the strong vibration and flame emission during the operation of the equipment, it is difficult to measure the multi-physical field by the traditional measurement technology. Tomographic laser absorption reconstruction based on laser absorption spectroscopy is a promising diagnostic for mapping the temperature and concentration images because it reduces the need for optical access. Since the inherently ill-posedness of the tomographic inverse problem leads to artifacts, low accuracy, poor robustness, and inability to visualize dynamic flow boundaries in the measured temperature and concentration images, the reconstructed algorithm must contain additional information to promote the presumed solution attributes. To address these problems, we propose a novel algorithm called Gaussian parametric level set method coupled with regularized Landweber to obtain the posterior topological structure of the turbulent reaction flow, which can compensate for the lack of a priori knowledge of the tomographic inverse problem. Numerical validation and algorithm comparison are performed using absorption spectra of water vapor at 7153.748 cm −1 and 7154.353 cm −1 and multi-angle parallel beam arrangement. The proposed approach significantly improves reconstructed quality and has excellent robustness. Even if the sparsely sampled data contains 10% random noise, the root mean square error is still within 0.12 in the temperature range 296–1600 K, approximately 50% of the error of the classical tomography methods. And the standard deviation is about 10% of the reconstructed results of Landweber. The proposed new method can obtain clear dynamic flow field boundaries, eliminate artifacts, improve measurement accuracy and robustness, and its applicability in harsh environments will benefit the supervision of combustion equipment, fuel utilization, pollutant control and optimal design of new combustion systems.