US 2026/0243161 A1 · WELL MASTERING

Abstract

A method including providing a machine learning model with access to a first well data source and a second well data source. The first well data source and the second well data source each includes one or more sets of data. Each record is associated with a single wellbore. The method including generating a first pair between a record of the first well data source and a record of the second well data source. The method includes displaying the record of the first well data source and the record of the second well data source and receiving a confirmation that the record of the first well data source and the record of the second well data source describe a real world wellbore. Then, training the machine learning model based on the confirmation and the record of the first well data source and the record of the second well data source.

Claims20

Dependent claims

A method comprising:

providing a machine learning model with access to a first well data source and a second well data source, wherein the first well data source and the second well data source each includes one or more records, wherein each record is associated with a single wellbore;
generating a first pair between a record of the first well data source and a record of the second well data source;
displaying the record of the first well data source and the record of the second well data source;
receiving a confirmation from a user that the record of the first well data source and the record of the second well data source describe a same real world wellbore;
and
training the machine learning model based on the confirmation and the record of the first well data source and the record of the second well data source.

3 more elements

A method comprising:

providing a machine learning model with access to a first well data source and a second well data source, wherein the first well data source and the second well data source each includes one or more sets of data, wherein each record is associated with a single wellbore;
blocking a record of the first well data source with a plurality of records of the second well data source to generate candidate pairs to create a block;
generating a similarity score for each candidate pair for one or more of a name attribute, a surface location attribute, a bottomhole location attribute, or a total wellbore depth attribute of the record of the first well data source and the candidate pair of the second well data source;
generating a probability score based on the similarity score for each candidate pair;
and
labeling a candidate pair as a match or nonmatch based on the probability score and a predetermined threshold.

3 more elements

Dependencies are inferred from the claim text.

Well mastering

Life of this patent

Well mastering

Life of this patent

Abstract

Claims20