US 11,668,853 B2
Petrophysical inversion with machine learning-based geologic priors
Amit Kushwaha, The Woodlands, TX (US); Ratnanabha Sain, Houston, TX (US); Jan Schmedes, Bellaire, TX (US); and Yunfei Yang, Palo Alto, CA (US)
Assigned to ExxonMobil Technology and Engineering Company, Spring, TX (US)
Filed by ExxonMobil Technology and Engineering Company, Spring, TX (US)
Filed on May 6, 2020, as Appl. No. 15/929,505.
Claims priority of provisional application 62/883,348, filed on Aug. 6, 2019.
Prior Publication US 2021/0041596 A1, Feb. 11, 2021
Int. Cl. G01V 99/00 (2009.01); G06F 30/27 (2020.01); E21B 49/00 (2006.01); G01V 1/28 (2006.01); G01V 1/30 (2006.01); G06N 3/08 (2023.01); G01V 11/00 (2006.01)
CPC G01V 99/005 (2013.01) [E21B 49/00 (2013.01); G01V 1/282 (2013.01); G01V 1/306 (2013.01); G01V 11/00 (2013.01); G06F 30/27 (2020.01); G06N 3/08 (2013.01); E21B 2200/20 (2020.05); G01V 2210/614 (2013.01); G01V 2210/6242 (2013.01); G01V 2210/6244 (2013.01); G01V 2210/667 (2013.01)] 26 Claims
OG exemplary drawing
 
1. A method for modeling a subsurface region, comprising:
obtaining a trained machine learning network, the trained machine learning network being trained at a frequency of training data that is matched to an expected frequency of input data to which the machine learning network is applied;
obtaining an initial petrophysical parameter estimate;
applying the trained machine learning network to the initial petrophysical parameter estimate to predict a geologic prior model;
obtaining geophysical data for the subsurface region;
obtaining geophysical parameters for the subsurface region; and
performing a petrophysical inversion with the geologic prior model, geophysical data, and geophysical parameters to generate a rock type probability model and an updated petrophysical parameter estimate;
wherein each one of (i) applying the trained machine learning network and (ii) performing the petrophysical inversion is carried out using a geophysical data analysis system.