Research on Tomographic Detection Performance in Methane Distribution Mapping: The Optimization of Evaluation Techniques and Reconstruction Algorithms

Methane leaks are common environmental issues that pose significant risks to public safety and the environment. Gas distribution mapping (GDM) can not only detect methane leaks but also evaluate the extent and locate the sources, with tomographic approach offering the advantage of noninterference and higher precision. However, existing methods have not extensively analyzed the effectiveness of the evaluation methods, leading to inaccuracies in the algorithm assessment process. This study proposes a mobile device based on the noncooperative-target tunable diode laser absorption spectroscopy (TDLAS) technology and designed a detailed method using Cramér-Rao bound (CRB) for the performance analysis of different GDM evaluation processes. The experiment results indicate that using a combination of maximum mean discrepancy (MMD) and Jensen-Shannon divergence (JSD) as evaluation methods is optimal. On this basis, the least Kullback-Leibler (KL) divergence objective and a choice of relative grid density of 1.5 is determined to be optimal. These findings will guide the further development of detection methods and expand the application capabilities of GDM technology.