Learning to reconstruct the bubble distribution with conductivity maps using Invertible Neural Networks and Error Diffusion

Electrolysis is crucial for eco-friendly hydrogen production, but gas bubbles generated during the process hinder reactions, reduce cell efficiency, and increase energy consumption. Additionally, these gas bubbles cause changes in the conductivity inside the cell, resulting in corresponding variations in the induced magnetic field around the cell. Therefore, measuring these gas bubble-induced magnetic field fluctuations using external magnetic sensors and solving the inverse problem of Biot-Savart Law allows for estimating the conductivity in the cell and, thus, bubble size and location. However, determining high-resolution conductivity maps from only a few induced magnetic field measurements is an ill-posed inverse problem. To overcome this, we exploit Invertible Neural Networks (INNs) to reconstruct the conductivity field. Our qualitative results and quantitative evaluation using random error diffusion show that INN achieves far superior performance compared to Tikhonov regularization.

MHT-X: Offline Multiple Hypothesis Tracking with Algorithm X

An efficient and versatile implementation of offline multiple hypothesis tracking with Algorithm X for optimal association search was developed using Python. The code is intended for scientific applications that do not require online processing. Directed graph framework is used and multiple scans with progressively increasing time window width are used for edge construction for maximum likelihood trajectories. The current version of the code was developed for applications in multiphase hydrodynamics, e.g. bubble and particle tracking, and is capable of resolving object motion, merges and splits. Feasible object associations and trajectory graph edge likelihoods are determined using weak mass and momentum conservation laws translated to statistical functions for object properties. The code is compatible with n-dimensional motion with arbitrarily many tracked object properties. This framework is easily extendable beyond the present application by replacing the currently used heuristics with ones more appropriate for the problem at hand. The code is open-source and will be continuously developed further.