Virtual ESP model to detect system degradation for preventive maintenance

1 . A method comprising:

selecting at least one wellbore variable of a wellbore operation;

for each of two or more machine learning models, training, via a processor, each respective machine learning model with a respective set of training data values of the at least one wellbore variable detected in a respective different time interval;

for each of the two or more machine learning models, processing, via the processor, for each of the at least one wellbore variables, a set of data samples of the respective at least one wellbore variable using each of the two or more trained machine learning models to output a respective model output response for each data sample of the set of data samples;

identifying a change in the wellbore operation, wherein identifying includes comparing the model output responses from two or more of the trained machine learning models; and

modifying the wellbore operation based on the identified change.

2 . The method of claim 1 , wherein the training each respective machine learning model includes

training a first machine learning model for each of the at least one wellbore variables with a first set of training data values of the at least one wellbore variable detected in a first time interval; and

training a second machine learning model for each of the at least one wellbore variables with a second set of training data values of the at least one wellbore variable detected in a second time interval;

and wherein the processing the set of data samples includes

processing, for each of the at least one wellbore variables, a set of data samples for the at least one wellbore variable using the first trained machine learning model to output a first model output response for each data sample of a set of data samples; and

processing, for each of the at least one wellbore variables, the set of data samples for the at least one wellbore variable using the second trained machine learning model to output a second model output response for each data sample of the set of data samples.

3 . The method of claim 2 , wherein identifying a change further includes:

detecting that the at least one wellbore variable has changed between the first time interval and the second time interval based on a difference between the first model output response and the second model output response.

4 . The method of claim 3 , further comprising modifying the wellbore operation based on detecting that the at least one wellbore variable has changed.

5 . The method of claim 3 , further comprising determining that at least one of the following has occurred,

equipment that is still operational and part of the wellbore operation has degraded beyond a degradable threshold based on detecting that the at least one wellbore variable has changed, or

determining that an event has occurred in the wellbore operation using a signature pattern for the event based on detecting that the at least one wellbore variable has changed.

6 . The method of claim 3 , wherein detecting that the at least one wellbore variable has changed comprises detecting that the at least one wellbore variable has changed in response the difference between the first model output response and the second model output response being beyond a change range.

7 . The method of claim 2 , further comprising:

determining the difference between the first model output response and the second model output response.

8 . The method of claim 7 , further comprising:

statistically comparing the difference between the first model output response and the second model output response.

9 . The method of claim 8 , wherein statistically comparing the difference comprises,

performing a paired random block design analysis for a pair of the first model output response and the second model output response for each data sample; and

performing a paired t-test for each pair of the paired block design analysis.

10 . The method of claim 8 , further comprising:

identifying which of the wellbore variables has changed between the first time interval and the second time interval based on the statistically comparing.

11 . The method of claim 10 , wherein identifying which of the wellbore variables has changed between the first time interval and the second time interval is based on a probability level of accuracy of the first trained machine learning model and the second trained machine learning model.

12 . The method of claim 2 , further comprising:

determining whether more than one of the at least one wellbore variables has changed between the first time interval and the second time interval; and

in response to determining that more than one of the at least one wellbore variables has changed between the first time interval and the second time interval,

determining whether there is at least one wellbore parameter common among the at least one wellbore variables; and

identifying the at least one wellbore parameter as a root cause of the change between the first time interval and the second time interval.

13 . The method of claim 2 , further comprising:

training wellbore parameters using at least one of the first or second machine learning models based on real time data from a number of sensors at the wellbore operation;

in response to at least one sensor of the number of sensors failing in the wellbore operation,

filling in the data from the at least one failed sensor using the associated trained wellbore parameters; and

creating a hybrid monitoring system for the wellbore operation using the real time data from the number of sensors and the trained wellbore parameters generated from the at least one of the first or second machine learning models.

14 . A non-transitory, computer-readable medium having instructions stored thereon that are executable by a processor, the instructions comprising:

instructions to select at least one wellbore variable of a wellbore operation;

for each of two or more machine learning models, instructions to train each respective machine learning model with a respective set of training data values of the at least one wellbore variable detected in a respective different time interval;

for each of the two or more machine learning models, instructions to process, for each of the at least one wellbore variables, a set of data samples of the respective at least one wellbore variable using each of the two or more trained machine learning models to output a respective model output response for each data sample of a set of data samples;

instructions to identify a change in the wellbore operation, wherein identifying includes comparing the model output responses from two or more of the trained machine learning models; and

instructions to modify the wellbore operation based on the identified change.

15 . The non-transitory, computer-readable medium of claim 14 , wherein the instructions to train each respective machine learning model includes

instructions to train a first machine learning model for each of the at least one wellbore variables with a first set of training data values of the at least one wellbore variable detected in a first time interval; and

instructions to train a second machine learning model for each of the at least one wellbore variables with a second set of train data values of the at least one wellbore variable detected in a second time interval;

and wherein the instructions to process the set of data samples includes

instructions to process, for each of the at least one wellbore variables, a set of data samples for the at least one wellbore variable using the first trained machine learning model to output a first model output response for each data sample of a set of data samples; and

instructions to process, for each of the at least one wellbore variables, the set of data samples for the at least one wellbore variable using the second trained machine learning model to output a second model output response for each data sample of the set of data samples.

16 . The non-transitory, computer-readable medium of claim 15 , wherein the instructions comprise instructions to detect that the at least one wellbore variable has changed between the first time interval and the second time interval based on a difference between the first model output response and the second model output response.

17 . The non-transitory, computer-readable medium of claim 16 , wherein the instructions comprise instructions to modify the wellbore operation based on detecting that the at least one wellbore variable has changed.

18 . The non-transitory, computer-readable medium of claim 16 , wherein the instructions comprise instructions to determine that equipment that is still operational and part of the wellbore operation has degraded beyond a degradable threshold based on detecting that the at least one wellbore variable has changed.

19 . The non-transitory, computer-readable medium of claim 16 , wherein the instructions to detect that the at least one wellbore variable has changed comprises instructions to detect that the at least one wellbore variable has changed in response the difference between the first model output response and the second model output response being beyond a change range.

20 . The non-transitory, computer-readable medium of claim 15 , wherein the instructions comprise,

instructions to determine the difference between the first model output response and the second model output response;

instructions to statistically compare the difference between the first model output response and the second model output response; and

instructions to identify which of the at least one wellbore variable has changed between the first time interval and the second time interval based on the statistically comparing.

21 . The non-transitory, computer-readable medium of claim 15 , wherein the instructions comprise,

instructions to determine whether more than one of the at least one wellbore variables has changed between the first time interval and the second time interval; and

in response to determining that more than one of the at least one wellbore variables has changed between the first time interval and the second time interval,

instructions to determine whether there is at least one wellbore parameter common among the at least one wellbore variable; and

instructions to identify the at least one wellbore parameter as a root cause of the change between the first time interval and the second time interval.

22 . A wellbore system comprising:

at least one sensor to be positioned at the surface of a wellbore and downhole in the wellbore;

a processor; and

a computer-readable medium having instructions stored thereon that are executable by

the processor to cause the processor to,

select at least one wellbore variable of a wellbore operation;

for each of two or more machine learning models, train, via the processor, each respective machine learning model with a respective set of training data values of the at least one wellbore variable detected in a respective different time interval;

for each of the two or more machine learning models, process, train, via the processor, for each of the at least one wellbore variables, a set of data samples of the respective at least one wellbore variable using each of the two or more trained machine learning models to output a respective model output response for

identify a change in the wellbore operation, wherein identifying includes comparing the model output responses from two or more of the trained machine learning models; and

modify the wellbore operation based on the identified change.

23 . The wellbore system of claim 22 , wherein the instructions to cause the processor to train each respective machine learning model include

instructions to cause the processor to train a first machine learning model for each of the at least one wellbore variables with a first set of training data values of the at least one wellbore variable detected in a first time interval; and

instructions to cause the processor to train a second machine learning model for each of the at least one wellbore variables with a second set of train data values of the at least one wellbore variable detected in a second time interval;

and wherein the instructions to cause the processor to process the set of data samples include

instructions to cause the processor to process, for each of the at least one wellbore variables, a set of data samples for the at least one wellbore variable using the first trained machine learning model to output a first model output response for each data sample of a set of data samples; and

instructions to cause the processor to process, for each of the at least one wellbore variables, the set of data samples for the at least one wellbore variable using the second trained machine learning model to output a second model output response for each data sample of the set of data samples.

24 . The wellbore system of claim 23 , wherein the instructions comprise instructions executable by the processor to cause the processor to detect that the at least one wellbore variable has changed between the first time interval and the second time interval based on a difference between the first model output response and the second model output response.

25 . The wellbore system of claim 24 , wherein the instructions comprise instructions executable by the processor to cause the processor to determine that equipment that is still operational and part of the wellbore operation has degraded beyond a degradable threshold based on detection that the at least one wellbore variable has changed.

26 . The wellbore system of claim 24 , wherein the instructions comprise instructions executable by the processor to cause the processor to modify the wellbore operation based on detection that the at least one wellbore variable has changed.

27 . The wellbore system of claim 23 , wherein the instructions comprise instructions executable by the processor to cause the processor to

determine the difference between the first model output response and the second model output response;

statistically compare the difference between the first model output response and the second model output response; and

identify which of the at least one wellbore variable has changed between the first time interval and the second time interval based on the statistically comparison.

28 . The wellbore system of claim 23 , wherein the instructions comprise instructions executable by the processor to cause the processor to,

determine whether more than one of the at least one wellbore variables has changed between the first time interval and the second time interval; and

in response to determining that more than one of the at least one wellbore variables has changed between the first time interval and the second time interval,

determine whether there is at least one wellbore parameter common among the at least one wellbore variable; and

identify the at least one wellbore parameter as a root cause of the change between the first time interval and the second time interval.