GEORLEARN

Objective:

The use of renewable hydrogen as green fuel and energy storage was deemed key to achieve the European Green Deal. However, its large-scale storage is still facing significant challenges. Measurement inversion via deep learning (DL) is a state-of-the-art approach used for underground storage-site detection and monitoring. However: 1) It requires a huge amount of training data; 2) DL training is expensive, and 3) There are no efficient and reliable DL techniques for multiscale electromagnetic measurement inversion. The goal of GEOLEARN is to guide hydrogen storage technologies by inverting subsurface multiscale electromagnetic measurements in real time using energy-efficient DL methods. For this purpose, GEOLEARN will leverage mixed-precision (MP) computations to maximise energy- and cost-efficiency, and ensure scalability. GEOLEARN proposes to address the above challenges as follows: 1) Develop MP finite element methods (FEMs) that can rapidly generate large training data; 2) Design MP DL algorithms that can efficiently process huge databases during training and invert measurements in real time, and 3) Apply the new techniques to invert multiscale geophysical electromagnetic measurements and guide hydrogen storage. We will collaborate with industry to disseminate the project results and maximise exploitation, and the new methods will lead to high impacts in and outside academia. The host has extensive experience in DL methods for inverse problems in geophysics and FEMs, and already collaborates with relevant companies. The secondment host is expert in high-performance computing and FEMs, and the applicant is expert in MP methods for scientific computing. This multidisciplinary research team is essential for the success of GEOLEARN, and will enhance the applicant's knowledge, network and skills, promoting his future career in research in Europe. The hosts and applicant will mutually benefit from the project outcomes and the industrial and academic collaborations.