Hybrid AI well modeling: Combining physics with data-driven methods
Mathilde is half way through her PhD project on well modeling where she combines mechanistic models founded on first order principals with general mathematical constructs that are particularly suited for adapting to production data. Today she is presenting her work so far in our Friday knowledge-sharing meetups. By combining these techniques, she strikes a good compromise between the utilization of a priori domain specific knowledge and data, not being over reliant on any of them. Her results are promising and the approach fits very well into the wider Solution Seeker model machinery.