PHYSICS-DRIVEN DEEP LEARNING INVERSION TO FLUID FLOW SIMULATORS

摘要

A method for a physics-driven deep learning-based inversion (600, 706) coupled to fluid flow simulators (160) may include obtaining measured data for a subsurface region, obtaining prior subsurface data for the subsurface region, and obtaining a physics-driven standard regularized joint inversion (350, 602) for at least two model parameters. The method may further include obtaining a case-based deep learning inversion (500, 604) characterized by a contracting path (402) and an expansive path (404). The method may further include forming the physics-driven deep learning inversion (600, 706) with the physics-driven standard regularized joint inversion (350, 602), the case-based deep learning inversion (500, 604), and a coupling operator (606) based on a penalty function. The method may further include forming a feedback loop (610) between the physics-driven standard regularized joint inversion (350, 602) and the case-based deep learning inversion (500, 604) for re-training the case-based deep learning inversion (612). The method may further include generating an inversion solution (1012) for reservoir monitoring.