DETAILED ACTION
This office action is in response to communication filed on February 26, 2026.
Response to Amendment
Amendments filed on February 26, 2026 have been entered.
The drawings have been amended.
The specification has been amended.
Claims 1, 3-7, 9-11, 14, 16 and 18-20 have been amended.
Claims 2, 8 and 12-13 have been canceled.
Claims 21-23 have been added.
Claims 1, 3-7, 9-11 and 14-23 have been examined.
Response to Arguments
Applicant’s arguments, see Remarks (p. 11), filed on 02/26/2026, with respect to the objections to the drawings have been fully considered. In view of the amendments to the drawings and the specification addressing the informalities raised in the previous office action, the objections to the drawings have been withdrawn.
Applicant’s arguments, see Remarks (p. 12), filed on 02/26/2026, with respect to the objections to the specification have been fully considered. In view of the amendments to the specification addressing the informalities raised in the previous office action, the objections to the specification have been withdrawn.
Applicant’s arguments, see Remarks (p. 12), filed on 02/26/2026, with respect to the objections to the claims have been fully considered. In view of the amendments to the claims addressing the informalities raised in the previous office action, the objections to the claims have been withdrawn. However, upon further consideration, new objections to the claims are presented to address additional informalities introduced by the current amendments.
Applicant’s arguments, see Remarks (p. 12), filed on 02/26/2026, with respect to the rejection of claim 20 under 35 U.S.C. 101 because the claimed invention is directed to non-statutory subject matter have been fully considered. In view of the amendments to the claim addressing the issues raised in the previous office action, the rejection of claim 20 has been withdrawn.
Applicant’s arguments, see Remarks (p. 12-18), filed on 02/26/2026, with respect to the rejection of claims 1, 3-7, 9-11 and 14-19 under 35 U.S.C. 101 because the claimed invention is directed to a judicial exception without significantly more have been fully considered. In view of the amendments to the claims addressing the issues raised in the previous office action, the rejections of claims 1, 3-7, 9-11 and 14-19 have been withdrawn.
Applicant argues (p. 16) that claim 1 has been amended to recite, in part, “deploying a plurality of new instances of the new trained machine learning model to the plurality of field sites; generating, at a field site of the plurality of field sites and from a new instance of the plurality of new instances of the new trained machine learning model, a pump output; and controlling operation of a pump at the field site of the plurality of field sites using the pump output, wherein controlling the operation of the pump includes adjusting a choke of the pump.” Applicant respectfully submits that such claim elements are integrated into a practical application. The Application explains that “As explained, a trained ML model may be utilized to process pump equipment data to classify and/or predict one or more issues (e.g., as behaviors, conditions, etc.). As an example, local controller may provide for control using output from a trained ML model. For example, where gas lock is detected via a trained ML model by classification and/or prediction using pump equipment data, a controller may control the pump equipment according to a specialized process that aims to alleviate gas lock such that normal pump operation can be achieved. As to a condition such as excessive rod stretch, a controller may issue a repair instruction and/or an adjustment instruction such that forces are reduced in an effort to minimize further stretch.” Application [00115]. Thus, controlling operation of a pump is an integration into a practical application.
These arguments are persuasive (see Examiner’s Note section for complete analysis).
Applicant’s arguments, see Remarks (p. 18-19), filed on 02/26/2026, with respect to the rejection of claims 1, 3-7, 9-11 and 14-20 under 35 U.S.C. 103 have been fully considered but are moot in view of new grounds of rejection.
Applicant argues (p. 18-19) that Watson appears to be silent regarding “assessing the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model,” as recited in amended claim 1 … Saghir appears to be silent regarding “assessing the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model,” as recited in amended claim 1.
These arguments are not persuasive.
The examiner submits that, under the broadest reasonable interpretation in light of the specification, Watson discloses the argued features by describing a system that can be applied in an oil field having multiple wells (see [0036]-[0037], [0040]) and used for identifying deviations resulting from drift of process parameters (see [0031]) in the applied process models (see [0062], [0089]-[0093], [0099]-[0100], [0108]; see rejection below for additional details).
Furthermore, the examiner submits that “Prior art is not limited just to the references being applied, but includes the understanding of one of ordinary skill in the art. The prior art reference (or references when combined) need not teach or suggest all the claim limitations, however, Office personnel must explain why the difference(s) between the prior art and the claimed invention would have been obvious to one of ordinary skill in the art. The “mere existence of differences between the prior art and an invention does not establish the invention’s nonobviousness.” Dann v. Johnston, 425 U.S. 219, 230, 189 USPQ 257, 261 (1976). The gap between the prior art and the claimed invention may not be “so great as to render the [claim] nonobvious to one reasonably skilled in the art.” Id. In determining obviousness, neither the particular motivation to make the claimed invention nor the problem the inventor is solving controls. The proper analysis is whether the claimed invention would have been obvious to one of ordinary skill in the art after consideration of all the facts. See 35 U.S.C. 103 or pre-AIA 35 U.S.C. 103(a). Factors other than the disclosures of the cited prior art may provide a basis for concluding that it would have been obvious to one of ordinary skill in the art to bridge the gap” (MPEP 2141, section III).
Drawings
The drawings were received on 02/26/2026. These drawings are acceptable.
Claim Objections
Claim 10 is objected to because of the following informalities: .
Claim language should read “The method of claim 3, comprising testing the containerized image prior to deploying the containerized image [[in]]as the plurality of new instances” in order to provide appropriate antecedence basis (e.g., see claim 3).
Appropriate correction is required.
Claim 11 is objected to because of the following informalities: .
Claim language should read “The method of claim 10, comprising, based on the testing, deciding to deploy the containerized image [[in]]as the plurality of new instances” in order to provide appropriate antecedence basis (e.g., see claim 3).
Appropriate correction is required.
Claim 18 is objected to because of the following informalities: .
Claim language should read “The method of claim 17, comprising, responsive to a notification generated by the web application, deploying the containerized image [[in]]as the plurality of new instances” in order to correct for minor informalities (e.g., see claim 3).
Appropriate correction is required.
Examiner’s Note
Claims 1, 3-7, 9-11 and 14-23 were evaluated for patent eligibility under 35 U.S.C. 101 using the SUBJECT MATTER ELIGIBILITY TEST FOR PRODUCTS AND PROCESSES described in the 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence (see also 2019 Revised Patent Subject Matter Eligibility Guidance) to determine patent eligibility under 35 U.S.C. 101.
Regarding claim 1, the examiner submits that under Step 1 of the 2024 Guidance Update on Patent Subject Matter Eligibility, Including on Artificial Intelligence (see also 2019 Revised Patent Subject Matter Eligibility Guidance) for evaluating claims for eligibility under 35 U.S.C. 101, the claim is to a process, which is one of the statutory categories of invention.
Continuing with the analysis, under Step 2A - Prong One of the test (see italic text for abstract idea), the claim recites:
“A method comprising:
receiving a plurality of results from a plurality of field devices, each of the plurality of field devices located at a respective field site of a plurality of field sites, wherein the plurality of results are generated using a plurality of instances of a trained machine learning model installed at the plurality of field sites;
receiving a plurality of sets of real-time field equipment data from field equipment at the plurality of field sites;
assessing the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model;
identifying, using the drift, a performance-related issue of the trained machine learning model;
responsive to the performance-related issue, identifying, from the plurality of results, new field data that have an impact on the trained machine learning model;
updating training data with the new field data to address the performance-related issue;
retraining the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model;
deploying a plurality of new instances of the new trained machine learning model to the plurality of field sites;
generating, at a field site of the plurality of field sites and from a new instance of the plurality of new instances of the new trained machine learning model, a pump output; and
controlling operation of a pump at the field site of the plurality of field sites using the pump output, wherein controlling the operation of the pump includes adjusting a choke of the pump”
Under Step 2A - Prong One of the test, the examiner submits that the claimed invention includes features that, under its broadest reasonable interpretation in light of the specification, cover performance of the limitation using mental processes and/or mathematical concepts to manipulate collected data to obtain results and evaluate the results.
Therefore, the claim recites a judicial exception under Step 2A - Prong One of the test.
Furthermore, under Step 2A - Prong Two of the test, the additional elements recited in the claim (see non-italic text for additional elements):
generally link the use of the judicial exception to a particular technological environment or field of use (e.g., updating machine learning models when drifting, see MPEP 2106.05(h));
add extra-solution activities (e.g., mere data gathering, source/type of data to be manipulated) using elements recited at a high level of generality (e.g., a plurality of field devices, a plurality of field sites, field equipment at the plurality of field sites) (see MPEP 2106.05(g));
add the words “apply it” (or an equivalent) with the judicial exception, or mere instructions to implement an abstract idea on a computer, or merely uses a computer as a tool to perform an abstract idea (e.g., training/updating machine learning models; see MPEP 2106.05(f)); and
when considering the claim as a whole, integrate the judicial exception into a practical application by reflecting an improvement to other technology or technical field and/or by applying the judicial exception with, or by use of, a particular machine (e.g., controlling operation of a pump at the field site of the plurality of field sites using the pump output, wherein controlling the operation of the pump includes adjusting a choke of the pump) (see MPEP 2106.05(a)-(b)).
Therefore, these additional elements, when considered individually and in combination, integrate the judicial exception into a practical application. The claim, when considered as a whole, is eligible at Prong Two of the Revised Step 2A (see 2019 Revised Patent Subject Matter Eligibility Guidance – Revised Step 2A, see also MPEP 2106.04(d)).
Similarly, independent claims 19-20 are directed to patent eligible subject matter as explained above with regards to claim 1.
Regarding the dependent claims 3-7, 9-11, 14-18 and 21-23, they were found to be patent eligible under 35 U.S.C. 101 by incorporating the eligible subject matter of their corresponding independent claims.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 3-7, 9-11, 14-15 and 19-23 are rejected under 35 U.S.C. 103 as being unpatentable over Watson (US 20150148919 A1), hereinafter ‘Watson’, in view of Saghir (US 20220316314 A1), hereinafter ‘Saghir’, and in further view of Gooneratne (US 20200190959 A1), hereinafter ‘Gooneratne’.
Regarding claim 1.
Watson discloses:
A method ([0001], [0028]-[0031]: methods for analytic validation of results for diagnosing a process state for decision making using an AIMC (artificially intelligent model-based controller) system are presented (see also [0007] regarding process models needed to be accurately tuned), with the AIMC system including multiple intelligent agents (IA) being employed during the analysis (see [0036], [0047]-[0048])) comprising:
receiving a plurality of results from a plurality of field devices ([0076], [0087]: intelligent agents (field devices) include data and modeling agents in oil wells in an oil field (see [0036]-[0037], [0040]), which run process models based on process information and transmit the results to status agents in the oil wells, the process models being used for determining conditions of sucker rod pumps in the oil wells (see [0037]-[0038])), each of the plurality of field devices located at a respective field site (Figs. 1 and 12) of a plurality of field sites ([0040]: multiple applications of the AIMC system are incorporated in an oil field, which contains multiple wells (plurality of field sites) (see also [0036]-[0037], [0073]), wherein the plurality of results are generated using a plurality of instances of a trained machine learning model installed at the plurality of field sites ([0039]-[0040]: intelligent agents at the oil field (see also [0192], [0196]) use learning methods such as Bayesian Networks during analysis (see current application at [0141] regarding machine learning model including a Bayesian model) (see also [0048], [0072], [0152] regarding updating/tuning (training) the models));
receiving a plurality of sets of real-time field equipment data from field equipment at the plurality of field sites ([0080], [0152], [0193]: data and modeling agents receive equipment data for analysis including for updating/tuning the process models);
assessing the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model ([0096], [0099]-[0100], [0108]: deviations in the process models are identified by the status agents, the deviations resulting from drift of process parameters (see [0031], [0062], [0089]-[0093]));
identifying, using the drift, a performance-related issue of the trained machine learning model ([0126]-[0138]: process model parameter tuning agents identify the causes of the deviations, which may include changes of model parameters or well conditions (see also [0063], [0139]-[0142], [0150], [0152]));
responsive to the performance-related issue, identifying, from the plurality of results, new field data that have an impact on the trained machine learning model ([0152]: process model parameter tuning agents use real-time data to validate the tuning of the process models (see also [0126], [0133]-[0139], [0150], [0155]-[0178])));
generating, at a field site of the plurality of field sites and from a new instance of the plurality of new instances of the new trained machine learning model, a pump output ([0126]-[0138], [0197]: process model parameter tuning agents ensure that the process models accurately reflect the actual process (see also [0031], [0048], [0072], [0083], [0152]), the process models being used for controlling and optimizing conditions of sucker rod pumps in the oil wells (see [0037]-[0038])); and
controlling operation of a pump at the field site of the plurality of field sites using the pump output ([0037]-[0038], [0095], [0107], [0179]-[0182]: the process models being are used for control and optimization of sucker rod pumps in the oil wells (see also [0119], [0197])).
Watson does not explicitly disclose:
updating training data with the new field data to address the performance-related issue;
retraining the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model;
deploying a plurality of new instances of the new trained machine learning model to the plurality of field sites; and
wherein controlling the operation of the pump includes adjusting a choke of the pump.
Regarding “updating training data with the new field data to address the performance-related issue; retraining the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and deploying a plurality of new instances of the new trained machine learning model to the plurality of field sites”, Saghir teaches:
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy, and retrained with different data (analogous to updating training data with the new field data to address the performance-related issue) until accuracy is achieved for deployment at the well site (see also [0068], [0074] and [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir to update training data with the new field data to address the performance-related issue; to retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and to deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding “controlling the operation of the pump includes adjusting a choke of the pump”, Gooneratne teaches:
“The programming instructions can include determining a priority, or urgency, of the generated operating condition (such as a risk of a stuck pipe) and communicate the detected operating condition to an automation system configured to proactively prevent or recover the operating condition to normalized operations, and apply a ranking criteria to the operating condition based on a severity and probability of occurrence of the operating condition (such as stuck pipe, lost circulation, or well influx), the ranking criteria configured to determine an order of operations for prioritizing a next recommended task. The programming instructions can also include executing an action automatically to address (or solve) the operating condition or provide guidance to manually intervene to address the operating condition” ([0007]: monitoring rig activities include detecting operating conditions and executing actions to address the operating conditions (see [0014]-[0016], [0025], [0096], [0099]-[0101] regarding controlling valves (analogous to choke) or pumps)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir, and in further view of Gooneratne to incorporate the controlling of the operation of the pump including adjusting a choke of the pump, in order to provide precise control of flow rates and pressure for stable field site operations.
Regarding claim 3.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not disclose:
generating a containerized image of the new trained machine learning model for deployment to the field site as the plurality of new instances of the new trained machine learning model.
Saghir further teaches:
“FIG. 5 is a block diagram illustrating an exemplary architecture 500 that may be used to deploy ML models on the edge gateway 132 ... Another container, Edge Agent, checks hardware integrity and ensures all necessary processes are running seamlessly … Cloud-connectivity allows the edge gateway 132 to be remotely monitored and managed for maintenance purpose, as well as for tracking the effectiveness and accuracy of the ML models, and can also trigger Transfer Learning remotely, a process discussed later herein. In short, the ML model deployment architecture 500 depicted in FIG. 5 is applicable to any edge gateway that meets minimum hardware requirements, and the use of the containers 504 to deploy the ML models and other associated components makes interoperability with different hardware platforms seamless and easy to manage” ([0075]: containers (analogous to containerized image) are used for deploying ML models while ensuring processes are running seamlessly (see also [0074])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to generate a containerized image of the new trained machine learning model for deployment to the field site as the plurality of new instances of the new trained machine learning model, in order to facilitate interoperability with different hardware platforms, as discussed by Saghir ([0075]).
Regarding claim 4.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson further discloses:
the performance-related issue comprises one or more of model classification drift or model prediction drift ([0096], [0099], [0108]: deviations/drift in process model predictions are identified by the status agent (see also [0031], [0062], [0089]-[0090])).
Regarding claim 5.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not explicitly disclose:
retraining the machine learning model occurs responsive to updating the training data.
Saghir further teaches:
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy, and retrained with different data (analogous to updating the training data) until accuracy is achieved for deployment at the well site (see also [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to retrain the machine learning model responsive to updating the training data, in order to retrain the ML models using real-time data, resulting in more accurate applications, as discussed by Saghir ([0118]).
Regarding claim 6.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not disclose:
updating the training data includes assigning labels to the new field data.
Saghir further teaches:
“The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations” ([0058]: data is labeled for training/updating ML models (see also [0068], [0083], [0085], [0090])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to incorporate the updating of the training data including assigning labels to the new field data, in order to produce a highly accurate ML model, as discussed by Saghir ([0068]).
Regarding claim 7.
Watson in view of Saghir and Gooneratne discloses all the features of claim 6 as described above.
Watson does not disclose:
retraining the trained machine learning model includes retraining using the labels via supervised learning.
Saghir further teaches:
“The locally generated data, once labeled and processed, may be used to update and further train the ML models, commission new edge devices, and generally improve well site operations” ([0058]: data is labeled for training/updating ML models (see also [0083], [0085], [0090])); and
“Training may be done using a self-supervised learning method in which labeled historical data is applied as an input to the ML models. Labeling refers to the process of annotating or describing the dynagraphs or specific regions in each dynagraph and creating a label or tag for those regions” ([0068]: training is done using supervised learning and labeled data).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to incorporate the retraining of the trained machine learning model including retraining using the labels via supervised learning, in order to produce a highly accurate ML model, as discussed by Saghir ([0068]).
Regarding claim 9.
Watson in view of Saghir and Gooneratne discloses all the features of claim 3 as described above.
Watson does not disclose:
evaluating the new trained machine learning model prior to building the containerized image.
Saghir further teaches:
“FIG. 5 is a block diagram illustrating an exemplary architecture 500 that may be used to deploy ML models on the edge gateway 132 ... Another container, Edge Agent, checks hardware integrity and ensures all necessary processes are running seamlessly … Cloud-connectivity allows the edge gateway 132 to be remotely monitored and managed for maintenance purpose, as well as for tracking the effectiveness and accuracy of the ML models, and can also trigger Transfer Learning remotely, a process discussed later herein. In short, the ML model deployment architecture 500 depicted in FIG. 5 is applicable to any edge gateway that meets minimum hardware requirements, and the use of the containers 504 to deploy the ML models and other associated components makes interoperability with different hardware platforms seamless and easy to manage” ([0075]: containers (analogous to containerized image) are used for deploying ML models); and
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy and retrained with different data until accuracy is achieved for deployment at the well site (see also [0068], [0074] and [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to evaluate the new trained machine learning model prior to building the containerized image, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding claim 10.
Watson in view of Saghir and Gooneratne discloses all the features of claim 3 as described above.
Watson does not disclose:
testing the containerized image prior to deploying the containerized image in the plurality of new instances.
Saghir further teaches:
“FIG. 5 is a block diagram illustrating an exemplary architecture 500 that may be used to deploy ML models on the edge gateway 132 ... Another container, Edge Agent, checks hardware integrity and ensures all necessary processes are running seamlessly … Cloud-connectivity allows the edge gateway 132 to be remotely monitored and managed for maintenance purpose, as well as for tracking the effectiveness and accuracy of the ML models, and can also trigger Transfer Learning remotely, a process discussed later herein. In short, the ML model deployment architecture 500 depicted in FIG. 5 is applicable to any edge gateway that meets minimum hardware requirements, and the use of the containers 504 to deploy the ML models and other associated components makes interoperability with different hardware platforms seamless and easy to manage” ([0075]: containers (analogous to containerized image) are used for deploying ML models); and
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy and retrained with different data until accuracy is achieved for deployment at the well site (see also [0068], [0074] and [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to test the containerized image prior to deploying the containerized image in the plurality of new instances, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding claim 11.
Watson in view of Saghir and Gooneratne discloses all the features of claim 10 as described above.
Watson does not disclose:
based on the testing, deciding to deploy the containerized image in the plurality of new instances.
Saghir further teaches:
“After satisfactory results are achieved from the learning phase, the trained and verified ML models 224 can be deployed to the edge gateway 132” ([0074]: once trained and verified ML models achieve satisfactory results, they are deployed); and
“FIG. 5 is a block diagram illustrating an exemplary architecture 500 that may be used to deploy ML models on the edge gateway 132 ... Another container, Edge Agent, checks hardware integrity and ensures all necessary processes are running seamlessly … Cloud-connectivity allows the edge gateway 132 to be remotely monitored and managed for maintenance purpose, as well as for tracking the effectiveness and accuracy of the ML models, and can also trigger Transfer Learning remotely, a process discussed later herein. In short, the ML model deployment architecture 500 depicted in FIG. 5 is applicable to any edge gateway that meets minimum hardware requirements, and the use of the containers 504 to deploy the ML models and other associated components makes interoperability with different hardware platforms seamless and easy to manage” ([0075]: containers (analogous to containerized image) are used for deploying ML models);
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to, based on the testing, decide to deploy the containerized image in the plurality of new instances, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding claim 14.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not disclose:
the trained machine learning model processes at least a portion of the new field data as image data.
Saghir further teaches:
“The training data may be data points, or where the well site being monitored uses a rod pump assembly, training may be performed using images of dynagraphs or dynacards” ([0085]: images of dynagraphs of a rod pump assembly are used as training data (see also [0009], [0069], [0087])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to configure the trained machine learning model to process at least a portion of the new field data as image data, in order to implement efficient and accurate analysis in oilfield operations.
Regarding claim 15.
Watson in view of Saghir and Gooneratne discloses all the features of claim 14 as described above.
Watson does not explicitly disclose:
the image data comprise dynacard images.
Saghir teaches:
“The training data may be data points, or where the well site being monitored uses a rod pump assembly, training may be performed using images of dynagraphs or dynacards” ([0085]: images of dynagraphs of a rod pump assembly are used as training data (see also [0009], [0069], [0087])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to implement the image data as dynacard images, in order to implement efficient and accurate analysis of commonly collected data regarding rod pump assemblies operations.
Regarding claim 19.
Watson discloses:
A system ([0001], [0028]-[0031], [0197]: apparatus for analytic validation of results for diagnosing process state for decision making using an AIMC (artificially intelligent model-based controller) system are presented (see also [0007] regarding process models needed to be accurately tuned), with the AIMC system including multiple intelligent agents (IA) being employed during the analysis (see [0036], [0047]-[0048])) comprising:
a processor ([0197]: apparatus includes at least one processor);
memory accessible to the processor ([0197]: apparatus includes storage coupled to the at least one processor); and
processor-executable instructions stored in the memory to instruct the system ([0197]: apparatus includes storage with instructions to be executed by the at least one processor) to:
receive a plurality of results from a plurality of field devices ([0076], [0087]: intelligent agents (field devices) include data and modeling agents in oil wells in an oil field (see [0036]-[0037], [0040]), which run process models based on process information and transmit the results to status agents in the oil wells, the process models being used for determining conditions of sucker rod pumps in the oil wells (see [0037]-[0038])), each of the plurality of field devices located at a respective field site (Figs. 1 and 12) of a plurality of field sites ([0040]: multiple applications of the AIMC system are incorporated in an oil field, which contains multiple wells (plurality of field sites) (see also [0036]-[0037], [0073]), wherein the plurality of results are generated using a plurality of instances of a trained machine learning model installed at the plurality of field sites ([0039]-[0040]: intelligent agents at the oil field (see also [0192], [0196]) use learning methods such as Bayesian Networks during analysis (see current application at [0141] regarding machine learning model including a Bayesian model) (see also [0048], [0072], [0152] regarding updating/tuning (training) the models));
receive a plurality of sets of real-time field equipment data from field equipment at the plurality of field sites ([0080], [0152], [0193]: data and modeling agents receive equipment data for analysis including for updating/tuning the process models);
assess the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model ([0096], [0099]-[0100], [0108]: deviations in the process models are identified by the status agents, the deviations resulting from drift of process parameters (see [0031], [0062], [0089]-[0093]));
identify, using the drift, a performance-related issue of the trained machine learning model ([0126]-[0138]: process model parameter tuning agents identify the causes of the deviations, which may include changes of model parameters or well conditions (see also [0063], [0139]-[0142], [0150], [0152]));
responsive to the performance-related issue, identify, from the plurality of results, new field data that have an impact on the trained machine learning model ([0152]: process model parameter tuning agents use real-time data to validate the tuning of the process models (see also [0126], [0133]-[0139], [0150], [0155]-[0178])));
generate, at a field site of the plurality of field sites and from a new instance of the plurality of new instances of the new trained machine learning model, a pump output ([0126]-[0138], [0197]: process model parameter tuning agents ensure that the process models accurately reflect the actual process (see also [0031], [0048], [0072], [0083], [0152]), the process models being used for controlling and optimizing conditions of sucker rod pumps in the oil wells (see [0037]-[0038])); and
control operation of a pump at the field site of the plurality of field sites using the pump output ([0037]-[0038], [0095], [0107], [0179]-[0182]: the process models being are used for control and optimization of sucker rod pumps in the oil wells (see also [0119], [0197])).
Watson does not explicitly disclose:
update training data with the new field data to address the performance-related issue;
retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model;
deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites; and
wherein controlling the operation of the pump includes adjusting a choke of the pump.
Regarding “update training data with the new field data to address the performance-related issue; retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites”, Saghir teaches:
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy, and retrained with different data (analogous to updating training data with the new field data to address the performance-related issue) until accuracy is achieved for deployment at the well site (see also [0068], [0074] and [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir to update training data with the new field data to address the performance-related issue; to retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and to deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding “controlling the operation of the pump includes adjusting a choke of the pump”, Gooneratne teaches:
“The programming instructions can include determining a priority, or urgency, of the generated operating condition (such as a risk of a stuck pipe) and communicate the detected operating condition to an automation system configured to proactively prevent or recover the operating condition to normalized operations, and apply a ranking criteria to the operating condition based on a severity and probability of occurrence of the operating condition (such as stuck pipe, lost circulation, or well influx), the ranking criteria configured to determine an order of operations for prioritizing a next recommended task. The programming instructions can also include executing an action automatically to address (or solve) the operating condition or provide guidance to manually intervene to address the operating condition” ([0007]: monitoring rig activities include detecting operating conditions and executing actions to address the operating conditions (see [0014]-[0016], [0025], [0096], [0099]-[0101] regarding controlling valves (analogous to choke) or pumps)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir, and in further view of Gooneratne to incorporate the controlling of the operation of the pump including adjusting a choke of the pump, in order to provide precise control of flow rates and pressure for stable field site operations.
Regarding claim 20.
Watson discloses:
One or more non-transitory computer-readable storage media comprising processor-executable instructions to instruct a wellsite computing system ([0197]: a computer readable medium includes instructions to be executed by a processor (see also [0037] regarding application for sucker rod pumps in oil wells)) to:
receive a plurality of results from a plurality of field devices ([0076], [0087]: intelligent agents (field devices) include data and modeling agents in oil wells in an oil field (see [0036]-[0037], [0040]), which run process models based on process information and transmit the results to status agents in the oil wells, the process models being used for determining conditions of sucker rod pumps in the oil wells (see [0037]-[0038])), each of the plurality of field devices located at a respective field site (Figs. 1 and 12) of a plurality of field sites ([0040]: multiple applications of the AIMC system are incorporated in an oil field, which contains multiple wells (plurality of field sites) (see also [0036]-[0037], [0073]), wherein the plurality of results are generated using a plurality of instances of a trained machine learning model installed at the plurality of field sites ([0039]-[0040]: intelligent agents at the oil field (see also [0192], [0196]) use learning methods such as Bayesian Networks during analysis (see current application at [0141] regarding machine learning model including a Bayesian model) (see also [0048], [0072], [0152] regarding updating/tuning (training) the models));
receive a plurality of sets of real-time field equipment data from field equipment at the plurality of field sites ([0080], [0152], [0193]: data and modeling agents receive equipment data for analysis including for updating/tuning the process models);
assess the plurality of results to track drift in the trained machine learning model across the plurality of instances of the trained machine learning model ([0096], [0099]-[0100], [0108]: deviations in the process models are identified by the status agents, the deviations resulting from drift of process parameters (see [0031], [0062], [0089]-[0093]));
identify, using the drift, a performance-related issue of the trained machine learning model ([0126]-[0138]: process model parameter tuning agents identify the causes of the deviations, which may include changes of model parameters or well conditions (see also [0063], [0139]-[0142], [0150], [0152]));
responsive to the performance-related issue, identify, from the plurality of results, new field data that have an impact on the trained machine learning model ([0152]: process model parameter tuning agents use real-time data to validate the tuning of the process models (see also [0126], [0133]-[0139], [0150], [0155]-[0178])));
generate, at a field site of the plurality of field sites and from a new instance of the plurality of new instances of the new trained machine learning model, a pump output ([0126]-[0138], [0197]: process model parameter tuning agents ensure that the process models accurately reflect the actual process (see also [0031], [0048], [0072], [0083], [0152]), the process models being used for controlling and optimizing conditions of sucker rod pumps in the oil wells (see [0037]-[0038])); and
control operation of a pump at the field site of the plurality of field sites using the pump output ([0037]-[0038], [0095], [0107], [0179]-[0182]: the process models being are used for control and optimization of sucker rod pumps in the oil wells (see also [0119], [0197])).
Watson does not explicitly disclose:
update training data with the new field data to address the performance-related issue;
retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model;
deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites; and
wherein controlling the operation of the pump includes adjusting a choke of the pump.
Regarding “update training data with the new field data to address the performance-related issue; retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites”, Saghir teaches:
“A determination is made at 1108 to verify whether the trained models can produce sufficiently accurate results within a given QoS. Verification is preferably performed using different data from the training data. For a negative determination, the workflow 1100 returns to the model training at 1102 for additional training. If the determination is positive, the trained and verified models are deployed to the well site at 1110” ([0086]: trained models are verified for accuracy, and retrained with different data (analogous to updating training data with the new field data to address the performance-related issue) until accuracy is achieved for deployment at the well site (see also [0068], [0074] and [0118]; see also Watson at [0126]-[0127], [0129], [0152] regarding tuning/validating the model to accurately reflect the actual process)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir to update training data with the new field data to address the performance-related issue; to retrain the trained machine learning model using the training data updated with the new field data, resulting in a new trained machine learning model; and to deploy a plurality of new instances of the new trained machine learning model to the plurality of field sites, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Regarding “controlling the operation of the pump includes adjusting a choke of the pump”, Gooneratne teaches:
“The programming instructions can include determining a priority, or urgency, of the generated operating condition (such as a risk of a stuck pipe) and communicate the detected operating condition to an automation system configured to proactively prevent or recover the operating condition to normalized operations, and apply a ranking criteria to the operating condition based on a severity and probability of occurrence of the operating condition (such as stuck pipe, lost circulation, or well influx), the ranking criteria configured to determine an order of operations for prioritizing a next recommended task. The programming instructions can also include executing an action automatically to address (or solve) the operating condition or provide guidance to manually intervene to address the operating condition” ([0007]: monitoring rig activities include detecting operating conditions and executing actions to address the operating conditions (see [0014]-[0016], [0025], [0096], [0099]-[0101] regarding controlling valves (analogous to choke) or pumps)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir, and in further view of Gooneratne to incorporate the controlling of the operation of the pump including adjusting a choke of the pump, in order to provide precise control of flow rates and pressure for stable field site operations.
Regarding claim 21.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not explicitly disclose:
controlling the operation of the pump includes alleviating gas lock at the pump.
However, Watson teaches:
“Significant gas interference can result in fluid pound which stresses the rod string and equipment or be indicative of a gas lock or pumpoff condition necessitating a shutoff of the equipment to allow fluid levels in the well to rise sufficiently for pumping to resume” ([0119]: diagnostic agents (see [0117]) assess conditions related to gas lock, with the process models being used for control and optimization of sucker rod pumps in the oil wells (see [0037]-[0038], [0107], [0179]-[0182], [0197])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to incorporate the controlling of the operation of the pump including alleviating gas lock at the pump, in order to stabilize equipment operations and improve overall field site conditions.
Regarding claim 22.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson further discloses:
the drift includes the trained machine learning model no longer adequately representing what is happening at the plurality of field sites ([0031]: deviations in the process models result from the models not reflecting current operating conditions (see also [0062])).
Regarding claim 23.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not explicitly disclose:
identifying the new field data includes applying a sampling process to the plurality of sets of real-time field equipment data, wherein the sampling process includes tagging probabilities and non-tagging probabilities in the plurality of sets of real-time field equipment data.
However, Watson teaches:
“A pattern recognition mechanism is included. The preferred implementation is a time series pattern recognition that provides a probabilistic match of incoming data stream to known patterns from a database to assist the process state tracking by PGM … In addition, the pattern recognition includes a classification mechanism that can identify and classify recurrent patterns that are associated with process deviations, store these patterns and track their reoccurrence in conjunction with process variables to assist with manual diagnosis or a diagnosis based on machine learning techniques using the pattern and the data observations associated with its occurrence” ([0096]: the status agent uses probabilistic match of incoming data during identification of deviations (see also [0029]-[0030], [0033], [0129], [0150])).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne to incorporate the identifying of the new field data including applying a sampling process to the plurality of sets of real-time field equipment data, wherein the sampling process includes tagging probabilities and non-tagging probabilities in the plurality of sets of real-time field equipment data, in order to reduce time by focusing on information affecting the actual process to improve the analysis and the response.
Claims 16-18 are rejected under 35 U.S.C. 103 as being unpatentable over Watson, in view of Saghir and Gooneratne, and in further view of Balogun (US 20130173505 A1), hereinafter ‘Balogun’.
Regarding claim 16.
Watson in view of Saghir and Gooneratne discloses all the features of claim 1 as described above.
Watson does not disclose:
assessing the plurality of results includes assessing the plurality of results via a web application.
Balogun teaches:
“A method for artificial lift system analysis is disclosed. The method comprises providing production well information for a plurality of the production wells each being associated with an artificial lift system. Artificial lift system failure alerts for the plurality of production wells are received and processed on a computer. A relevance measure for each of the artificial lift system failure alerts is determined responsive to the production well information. A summary of the artificial lift system failure alerts is displayed in an ordering based on the relevance measure” ([0008]: artificial lift system failure alerts are generated using well information and machine learning models (see [0075]; see also [0013] and [0070] regarding retraining when alerts are incorrect)); and
“As will be described, the invention can be implemented in numerous ways, including for example as a method (including a computer-implemented method), a system (including a computer processing system), an apparatus, a computer readable medium, a computer program product, a graphical user interface, a web portal, or a data structure tangibly fixed in a computer readable memory” ([0097]: the method can be applied as a web portal (web application)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir and Gooneratne, and in further view of Balogun, to incorporate the assessing of the plurality of results including assessing the plurality of results via a web application, in order to easily and safely manage remote well sites.
Regarding claim 17.
Watson in view of Saghir, Gooneratne and Balogun discloses all the features of claim 16 as described above.
Watson does not disclose:
the web application comprises features for assessing performance of the trained machine learning model and for assessing performance of a containerized image of the new trained machine learning model.
Balogun further teaches:
“A method for artificial lift system analysis is disclosed. The method comprises providing production well information for a plurality of the production wells each being associated with an artificial lift system. Artificial lift system failure alerts for the plurality of production wells are received and processed on a computer. A relevance measure for each of the artificial lift system failure alerts is determined responsive to the production well information. A summary of the artificial lift system failure alerts is displayed in an ordering based on the relevance measure” ([0008]: artificial lift system failure alerts are generated using well information and machine learning models (see [0075]; see also [0013] and [0070] regarding retraining when alerts are incorrect (analogous to assessing performance))); and
“As will be described, the invention can be implemented in numerous ways, including for example as a method (including a computer-implemented method), a system (including a computer processing system), an apparatus, a computer readable medium, a computer program product, a graphical user interface, a web portal, or a data structure tangibly fixed in a computer readable memory” ([0097]: the method can be applied as a web portal (web application)).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir, Gooneratne and Balogun to implement the web application comprising features for assessing performance of the trained machine learning model and for assessing performance of a containerized image of the new trained machine learning model, in order to validate performance of the models while managing remote well sites.
Regarding claim 18.
Watson in view of Saghir, Gooneratne and Balogun discloses all the features of claim 17 as described above.
Watson does not disclose:
responsive to a notification generated by the web application, deploying the containerized image in the plurality of new instances.
Saghir further teaches:
“After satisfactory results are achieved from the learning phase, the trained and verified ML models 224 can be deployed to the edge gateway 132” ([0074]: once trained and verified ML models achieve satisfactory results, they are deployed (see also [0118])); and
“FIG. 5 is a block diagram illustrating an exemplary architecture 500 that may be used to deploy ML models on the edge gateway 132 ... Another container, Edge Agent, checks hardware integrity and ensures all necessary processes are running seamlessly … Cloud-connectivity allows the edge gateway 132 to be remotely monitored and managed for maintenance purpose, as well as for tracking the effectiveness and accuracy of the ML models, and can also trigger Transfer Learning remotely, a process discussed later herein. In short, the ML model deployment architecture 500 depicted in FIG. 5 is applicable to any edge gateway that meets minimum hardware requirements, and the use of the containers 504 to deploy the ML models and other associated components makes interoperability with different hardware platforms seamless and easy to manage” ([0075]: containers (analogous to containerized image) are used for deploying ML models).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Watson in view of Saghir, Gooneratne and Balogun to deploy the containerized image in the plurality of new instances, responsive to a notification generated by the web application, in order to ensure high level of confidence for the predictions through ML-based analytics, as discussed by Saghir ([0059]).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Vazquez; Manuel et al., US 6343656 B1¸ System and method for optimizing production from a rod-pumping system
Reference discloses determining optimum operating conditions of a rod-pumping system by evaluating its power consumption using a mathematical model.
WATSON; Jeff, US 20200201267 A1, METHOD AND APPARATUS FOR ARTIFICALLY INTELLIGENT MODEL-BASED CONTROL OF DYNAMIC PROCESSES USING PROBABILISTIC AGENTS
Reference discloses controlling oil field production process using intelligent agents to collect sensor data and apply probabilistic models for evaluating operating conditions and provide recommendations of appropriate actions.
Applicant’s amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to LINA CORDERO whose telephone number is (571)272-9969. The examiner can normally be reached 9:30 am - 6:00 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, ANDREW SCHECHTER can be reached at 571-272-2302. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/LINA CORDERO/Primary Examiner, Art Unit 2857
Prosecution Insights