A Machine Learning paper research on Petroleum Engineering

Alfonso R. Reyes
(25 March 2018)

Online


So happy to find that OnePetro has improved its search algorithms. Now, lowercase, uppercase and dashed keywords return the same results. This wasn’t the case few months ago. You can see the results below.

This is a search of papers using R, #rstats, and the package petro.One. You could see that “neural network”, “neural-network, and”NEURAL NETWORK“, return the same number of papers: 3197. The same with”data driven" and “data-driven”. This makes the paper search much more efficient and fast. Maybe a little tweaking more on the root of the words like in the case of “neural networks” vs. “neural network”; they should be the same (stem). But it is so close: a difference of 9 papers over 3197.

This is part of a research study I am doing to find how many artificial intelligence and machine learning applications have been performed and deployed in the petroleum industry. So far, I have found 100+ different applications across disciplines such as reservoir, production, well engineering, drilling; completions, intervention, logging; geophysics, geology, seismic; petrophysics, PVT, etc.

These are some of the results. I haven’t finished yet. I will publish it as part of an R package in few weeks.

I am classifying the applications, in this case, by the machine learning algorithm used.

There are other ML algorithms that have been used over the years.

What it is interesting is that some of those applications, use combination of of algorithms; sometimes, borrowing from the control theory field as well.


comments powered by Disqus