DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Claims 1-15 are presented for examination.
Continued Examination under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on July 6, 2026 has been entered.
Response to Amendment
Applicant’s amendments have obviated the remaining specification objections, as well as the rejections under 35 USC §§ 101, 112. Therefore, those objections and rejections are withdrawn.
Claim Rejections - 35 USC § 103
Claims 1-2, 4-6, 8, 10-12, and 14-15 are rejected under 35 U.S.C. 103 as being unpatentable over Heidari et al. (US 20210087925) (“Heidari”) in view of Chen et al. (US 20150211357) (“Chen”) and further in view of Ben Kimon et al. (US 20200210849) (“Ben Kimon”).
Regarding claim 1, Heidari discloses “[a] system for predicting …, the system comprising:
one or more physical processors configured by machine-readable instructions (Heidari paragraph 18 discloses that the method is performed by a processor coupled with a computer-readable storage medium containing instructions executed by the processor) to:
obtain well operation information, the well operation information characterizing operation characteristics of a well during production of oil and/or gas from the well for a duration of time (time-series data [for a duration of time] may be different types of data associated with a hydraulic fracturing well; each type of data may be encoded separately using an LSTM encoder; time-series data may include flow rate [of oil, i.e., the information is obtained during production], proppant concentration, fluid concentration, chemical additive concentration, and pressure [operation characteristics] – Heidari, paragraph 68);
determine reconstructed operation characteristics of the well for the duration of time using an unsupervised machine-learning model (time-series data may be different types of data associated with a hydraulic fracturing well; each type of data may be encoded separately using an LSTM encoder [unsupervised machine-learning model]; time-series data may include flow rate, proppant concentration, fluid concentration, chemical additive concentration, and pressure [operation characteristics]; the encoded data may be decoded [reconstructed] to retrieve the original time-series data – Heidari, paragraph 68), wherein the unsupervised machine-learning model is trained using historical well operation information, the historical well operation information characterizing the operation characteristics of the well for a period of time preceding the duration of time (hydraulic fracturing job optimization system can generate [train] a machine learning model based on historical [i.e., preceding the duration of time] production data [historical well operation information; operation characteristics include production as well as the above metrics] – Heidari, paragraph 78); and
predict … a future occurrence … at the well before the [occurrence] occurs, based on … the operation characteristics of the well for the duration of time and the reconstructed operation characteristics of the well for the duration of time (hydraulic fracturing job optimization system can determine an optimized job design for a hydraulic fracturing well using a prediction based on the machine learning model – Heidari, paragraph 79 [future occurrence = implementation of design that will lead to optimized performance]; optimization system can generate the model based on the time-series data; each type of data may be encoded separately using an LSTM encoder; time-series data may include flow rate, proppant concentration, fluid concentration, chemical additive concentration, and pressure [operation characteristics]; the encoded data may be decoded [to create reconstructed operation characteristics] to retrieve the original time-series data – id. at paragraph 68 [since the prediction is based on the ML model, and the ML model encodes and decodes the operation characteristics, the prediction is based on these characteristics]) ...;
wherein one or more preventative or mitigative actions are performed in response to the future occurrence … predicted by the system, the one or more preventative or mitigative actions including: changing a production rate of the well, performing an intervention on the well at a scheduled time, or controlling operations to prevent uncommanded shut-in of the well (hydraulic fracturing job optimization system can determine an optimized job design for a hydraulic fracturing well having an objective function using a prediction based on the machine learning model; the objective function may include maximizing cumulative barrel of oil equivalent for a first six months of production [i.e., controlling the production rate by changing it to the maximum cumulative barrel of oil equivalent, thereby preventing suboptimal production] – Heidari, paragraph 79).”
Heidari appears not to disclose explicitly the further limitations of the claim. However, Chen discloses “predict[ing], during production of oil and/or gas from the well, a future occurrence of an asphaltene anomaly…, wherein the asphaltene anomaly is a production of asphaltene in the well and/or an effect of asphaltene production in the well (based on known fluid parameters and asphaltene gradients determined at one or more depths in a well, an asphaltene onset pressure [asphaltene anomaly that causes the production of asphaltene] prediction can be made at an additional depth – Chen, paragraph 51; see also paragraphs 47 (characterizing the predictions as future-regarding), 49 (disclosing that a flow assurance problem can be identified through comparison of predicted and measured asphaltene onset pressures, i.e., the analysis is performed while hydrocarbons are flowing/being produced))….”
Chen and the instant application both relate to asphaltene prediction and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Heidari to predict the future occurrence of asphaltene-related events, as disclosed by Chen, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the user to take preventative measures before an anomaly occurs, thereby optimizing the well’s performance. See Chen, paragraph 51.
Neither Heidari nor Chen appears to disclose explicitly the further limitations of the claim. However, Ben Kimon discloses “predict[ing an] … anomaly … based on a difference between the … characteristics … for the duration of time and the reconstructed … characteristics … for the duration of time (reconstruction difference threshold for anomalies is determined by receiving a plurality of fraudulent transaction data, and generating a reconstruction difference [score based on a difference between characteristics and reconstructed characteristics] – Ben Kimon, paragraph 34) ….”
Ben Kimon and the instant application both relate to anomaly detection using autoencoders and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari and Chen to determine that an anomaly has occurred based on a reconstruction error threshold, as disclosed by Ben Kimon, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow for anomaly detection in the presence of noisy datasets, thereby preventing problems that require remediation. See Ben Kimon, paragraph 2.
Claim 11 is a method claim corresponding to system claim 1 and is rejected for the same reasons as given in the rejection of that claim.
Regarding claim 2, Heidari, as modified by Chen/Ben Kimon, discloses that “the unsupervised machine-learning model includes a linear unsupervised machine-learning model and/or a non-linear unsupervised machine-learning model (optimization system can generate the model based on the time-series data; each type of data may be encoded separately using an LSTM encoder; time-series data may include flow rate, proppant concentration, fluid concentration, chemical additive concentration, and pressure; the encoded data may be decoded to retrieve the original time-series data – Heidari, paragraph 68 [LSTM encoder/decoder = LSTM autoencoder, i.e., non-linear unsupervised machine learning model]).”
Claim 12 is a method claim corresponding to system claim 2 and is rejected for the same reasons as given in the rejection of that claim.
Regarding claim 4, Heidari, as modified by Chen/Ben Kimon, discloses that “the non-linear unsupervised machine-learning model includes a long short-term memory autoencoder (optimization system can generate the model based on the time-series data; each type of data may be encoded separately using an LSTM encoder; time-series data may include flow rate, proppant concentration, fluid concentration, chemical additive concentration, and pressure; the encoded data may be decoded to retrieve the original time-series data – Heidari, paragraph 68 [LSTM encoder/decoder = LSTM autoencoder, i.e., non-linear unsupervised machine learning model]).”
Claim 14 is a method claim corresponding to system claim 4 and is rejected for the same reasons as given in the rejection of that claim.
Regarding claim 5, the rejection of claim 1 is incorporated. Chen further discloses that “prediction of the future occurrence of the asphaltene anomaly at the well based on the operation characteristics of the well for the duration of time … includes:
determination of an anomaly score based on … the operation characteristics of the well for the duration of time (based on known fluid parameters and asphaltene gradients [operation characteristics of the well for a duration of time] determined at one or more depths in a well, an asphaltene onset pressure [anomaly score] prediction can be made at an additional depth – Chen, paragraph 51; see also paragraph 47 (characterizing the predictions as future-regarding)) … ; and
prediction of the future occurrence of the asphaltene anomaly at the well (based on known fluid parameters and asphaltene gradients determined at one or more depths in a well, an asphaltene onset pressure [asphaltene anomaly] prediction can be made at an additional depth – Chen, paragraph 51; see also paragraph 47 (characterizing the predictions as future-regarding)) ....” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified Heidari to predict asphaltene anomalies based on an anomaly score determined from well operation characteristics, as disclosed by Chen, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow the user to take preventative measures before an anomaly occurs, thereby optimizing the well’s performance. See Chen, paragraph 51.
Neither Heidari nor Chen appears to disclose explicitly the further limitations of the claim. However, Ben Kimon discloses that “prediction of the … occurrence … based on … the reconstructed … characteristics … for the duration of time includes:
determination of [a] … score based on a difference between the … characteristics … for the duration of time and the reconstructed … characteristics … for the duration of time (reconstruction difference threshold for anomalies is determined by receiving a plurality of fraudulent transaction data, and generating a reconstruction difference [score based on a difference between characteristics and reconstructed characteristics] – Ben Kimon, paragraph 34); and
prediction of the … occurrence … based on a comparison between the anomaly score and an anomaly score threshold (first reversion transaction is determined to be anomalous by comparing the reconstruction difference [score] with the reconstruction error threshold for anomaly [anomaly score threshold] – Ben Kimon, claim 5).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari and Chen to determine that an anomaly has occurred based on a reconstruction error threshold, as disclosed by Ben Kimon, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow for anomaly detection in the presence of noisy datasets, thereby preventing problems that require remediation. See Ben Kimon, paragraph 2.
Claim 15 is a method claim corresponding to system claim 5 and is rejected for the same reasons as given in the rejection of that claim.
Regarding claim 6, the rejection of claim 5 is incorporated. Heidari further discloses “operation characteristics of the well for the period of time”, as shown above in the rejection of claim 1.
Neither Heidari nor Chen appears to disclose explicitly the further limitations of the claim. However, Ben Kimon discloses that “the anomaly score threshold is determined based on the … characteristics … and reconstructed … characteristics (reconstruction difference threshold for anomalies is determined by receiving a plurality of fraudulent transaction data, and generating a reconstruction difference [difference between characteristics and reconstructed characteristics] – Ben Kimon, paragraph 34; reconstruction error threshold for fraud generator receives the reconstruction differences and determines a reconstruction error threshold for fraud (e.g., average of the reconstruction differences, weighted average of the reconstruction differences, etc.) – id. at paragraph 35) ….” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari and Chen to determine the anomaly threshold based on the characteristics and reconstructed characteristics, as disclosed by Ben Kimon, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow for anomaly detection in the presence of noisy datasets, thereby preventing problems that require remediation. See Ben Kimon, paragraph 2.
Regarding claim 8, Heidari, as modified by Chen and Ben Kimon, discloses that “the one or more physical processors are further configured by the machine-readable instructions to present a visualization of the comparison between the anomaly score and the anomaly score threshold (first reversion transaction is determined to be anomalous by comparing the reconstruction difference [score] with the reconstruction error threshold for anomaly [anomaly score threshold] – Ben Kimon, claim 5; toolbar application may display a user interface in connection with a browser application [including the comparison] – id. at paragraph 47).” It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari and Chen to display the difference between the score and the threshold, as disclosed by Ben Kimon, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would allow for anomaly detection in the presence of noisy datasets, thereby preventing problems that require remediation. See Ben Kimon, paragraph 2.
Regarding claim 10, Heidari, as modified by Chen/Ben Kimon, discloses that “the operation characteristics of the well include[] pressure and temperature at the well (Heidari paragraph 26 discloses that the data collected include pressure data and digital temperature sensing data).”
Claims 3 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Heidari in view of Chen and Ben Kimon and further in view of Yoon et al. (US 20200320402) (“Yoon”).
Regarding claim 3, neither Heidari, Ben Kimon, nor Chen appears to disclose explicitly the further limitations of the claim. However, Yoon discloses that “the linear unsupervised machine-learning model includes a principal component analysis algorithm (some unsupervised learning models focus on characterization of normality and detecting samples out of the normality, e.g., PCA for linearity – Yoon, paragraph 22).”
Yoon and the instant application both relate to PCA and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari, Ben Kimon, and Chen to use PCA as part of the machine learning algorithm, as disclosed by Yoon, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would save processor resources by reducing the dimensionality of the data. See Yoon, paragraph 22.
Claim 13 is a method claim corresponding to system claim 3 and is rejected for the same reasons as given in the rejection of that claim.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Heidari in view of Chen and further in view of Ben Kimon and Beggel et al. (US 20190130279) (“Beggel”).
Regarding claim 7, the rejection of claim 6 is incorporated. Heidari further discloses “operation characteristics of the well for the period of time”, as shown above in the rejection of claim 1.
Neither Heidari, Chen, nor Ben Kimon appears to disclose explicitly the further limitations of the claim. However, Beggel discloses that “the anomaly score threshold is determined based on the … characteristics … and reconstructed …characteristics … such that a threshold percentage of historical anomaly scores satisfies the anomaly score threshold (anomaly decision threshold can be chosen based on reconstruction error [i.e., difference between characteristics and reconstructed characteristics] and can depend on the distribution of reconstruction errors during training; several alternatives are available to determine this threshold, including a percentile of reconstruction errors, such that, e.g., only 5% of all training images exceed this reconstruction error [i.e., satisfy the threshold] – Beggel, paragraph 43).”
Beggel and the instant application both relate to anomaly detection using autoencoders and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari, Chen, and Ben Kimon to determine the anomaly score threshold based on a percentage, as disclosed by Beggel, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would provide an objective and simple-to-execute criterion for determining whether an anomaly has occurred, allowing the user to take corrective action based on the detected anomaly. See Beggel, paragraph 43.
Claim 9 is rejected under 35 U.S.C. 103 as being unpatentable over Heidari in view of Chen and Ben Kimon and further in view of Marcharet (US 20130254153) (“Marcharet”).
Regarding claim 9, the rejection of claim 1 is incorporated. Heidari further discloses that “the unsupervised machine-learning model is [trained] using the well operation information (hydraulic fracturing job optimization system can generate [train] a machine learning model based on historical production data [well operation information] – Heidari, paragraph 78).”
Neither Heidari, Ben Kimon, nor Chen appears to disclose explicitly the further limitations of the claim. However, Marcharet discloses that “the unsupervised machine-learning model is retrained (illustrated method performs unsupervised retraining of a classification model – Marcharet, paragraph 45) ….”
Marcharet and the instant application both relate to machine learning and are analogous. It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to have modified the combination of Heidari, Ben Kimon, and Chen to retrain the model in unsupervised fashion, as disclosed by Marcharet, and an ordinary artisan could reasonably expect to have done so successfully. Doing so would improve performance of the model with a particular use case with a distribution of unlabeled test data. See Marcharet, paragraph 24.
Response to Arguments
Applicant's arguments filed July 6, 2026 (“Remarks”) have been fully considered but they are, except insofar as rendered moot by the withdrawal of a ground of rejection, not persuasive.
Applicant argues that the Heidari/Chen/Ben Kimon operation does not disclose obtaining well operation information during production of oil and/or gas from the well because the data of Heidari relate to fracking time-series data and fracking is performed before production. Remarks at 13-14. However, paragraph 68 of Heidari explicitly discloses that the time-series data obtained include flow rate, which can only be measured while hydrocarbons are actually flowing. Applicant further argues that the combination does not disclose predicting a future occurrence of an asphaltene anomaly during production from the well because the asphaltene onset pressure of Chen is not a production of asphaltene at the well and the prediction in Heidari and Chen allegedly do not occur during production of oil and/or gas from the well. Remarks at 14-16. However, as to the first of these arguments, paragraph 1 of Chen discloses that the asphaltene onset pressure is the pressure at a given temperature at which asphaltene precipitation begins. Thus, a prediction of asphaltene onset pressure is a prediction of when asphaltene will begin to be produced. As to the second of these points, as shown in the updated rejection, paragraph 49 of Chen discloses that a flow assurance problem is identified through comparison of predicted and measured asphaltene onset pressures, and paragraph 32 discloses that testing is performed through the sampling of flowing fluids, i.e., of oil and gas. Since the amount of “production” is not claimed, any sampling of oil and gas, even for purposes of testing, may be deemed a variety of production under the broadest reasonable interpretation of the term “production” in light of the specification.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to RYAN C VAUGHN whose telephone number is (571)272-4849. The examiner can normally be reached M-R 7:00a-5:00p ET.
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, Kamran Afshar, can be reached at 571-272-7796. 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.
/RYAN C VAUGHN/ Primary Examiner, Art Unit 2125
Prosecution Insights