US 12,234,714 B2
Cleansing of drilling sensor readings
Adrian Marius Ambrus, Stavanger (NO); and Pradeepkumar Ashok, Austin, TX (US)
Assigned to Intellicess, Inc., Austin, TX (US)
Filed by Intellicess, Inc., Austin, TX (US)
Filed on Feb. 28, 2018, as Appl. No. 15/908,456.
Claims priority of provisional application 62/464,475, filed on Feb. 28, 2017.
Prior Publication US 2022/0284330 A1, Sep. 8, 2022
Int. Cl. E21B 44/00 (2006.01); G06F 18/15 (2023.01); G06N 7/01 (2023.01)
CPC E21B 44/00 (2013.01) [G06F 18/15 (2023.01); G06N 7/01 (2023.01); E21B 2200/20 (2020.05)] 20 Claims
OG exemplary drawing
 
1. A drilling monitoring system comprising:
a plurality of sensors that measure attributes of drilling equipment or well conditions in real time;
at least one data connection that transmits measurements made by the plurality of sensors to a control unit;
in the control unit, a computer processor executing machine readable code stored in a non-transitory medium to:
construct a prior Bayesian network model using previously received measurements;
synchronize the received measurements made by the plurality of sensors to correspond to a time at which each measurement was made;
synchronize the received measurements made by the plurality of sensors to correspond to a depth of the measurement;
identify a current rig activity associated with the drilling equipment using the received measurements as well as the time and the depth of the received measurements;
process received measurements to identify missing measurements and remove outlier measurements based, at least in part, on the identified rig activity;
create cleansed measurements by removing faulty measurements from the received measurements by comparing each of the received measurements to an accuracy and precision associated with the sensor making that measurement to the prior Bayesian network model to identify and remove measurements outside of a range defined by a lower bound and an upper bound for the sensor making the measurement and by comparing the received measurements to generated belief patterns for identifying faults;
repopulating the prior Bayesian network model using the cleansed data to allow for an updated Bayesian network model; and
providing the cleansed measurements for drilling analytics associated with a drilling operation related to the drilling equipment.