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 .
Response to Amendment
This Non-Final rejection is filed in response to Request for Continued Examination (RCE) filed 03/09/2026.
Claims 1, 10, 19, & 27 are amended.
Claims 1-11, 13-20, 22-24, & 26-28 remain pending.
Response to Arguments
Argument 1, applicant argues in Applicant Arguments/Remarks made in an Amendment filed 01/019/2026 pg. 8-10 that prior art fails to teach the primary claim limitation, “wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device”.
Response to Argument 1, Upon further examination of the prior art, the examiner respectfully disagrees. It is noted that regarding the definition maximum potential power generation mode, applicant’s specification at most merely cites that power generation devices may be operating below MPPG mode, in an MPPG mode, or outside an MPPG mode in para. [0041], “Applying a large weight to the distances of the power outputs 114 applied can highlight the operation of a particular power generation device 108 below the maximum potential power generation (MPPG) mode of the power generation device 108. For example, the set of power generation devices 108 with similar features 110 can include several power generation devices 108 that are operating in an MPPG mode and a one power generation device 108 that is operating outside of an MPPG mode”. This does not define what it means for a power generation device to operate in a mode for maximum power output as it appears that while in a MPPG mode a device may be operating below, at, or even outside its ability to generate a maximum potential power. Thus the claims read in light of the specification lead to a BRI for the limitation, “a maximum potential power generation (MPPG) mode”, to encompass any power generation device operating in a mode in which it is capable of outputting a maximum potential power threshold of some sort and that the measurements that may be taken may vary from below, at, or outside such a threshold.
The examiner notes that Granade teaches in para. [0039], “An Exception may refer to raw data indicative of values that are outside of a predetermined range. For example, a Voltage Exception may occur if a voltage level measured by a meter is outside of a predetermined range of acceptable levels.” Wherein the examiner notes a power grid may be operating in a mode where a maximum potential safe threshold output is possible, yet measurements outside such safe or acceptable maximum potential power output may be measured and found to have a difference above the safe threshold.
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.
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.
Claim(s) 1-2, 5-6, 9, 10-11, 13-15, 17, 19, 22-23, 26-27 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20200065677 “Lopez”, further light of U.S. Patent Application Publication NO. 20190025365 “Granade”, and further in light of U.S. Patent Application Publication NO. 20230214703 “Maheswari”.
Claim 1:
Lopez teaches a computer-implemented method, comprising: receiving, by a training data set generator engine (i.e. para. [0054], Pre-processing includes cleaning the data, which may entail removing outlier values and the like, and other modifications to the data to increase the efficiency of training (e.g., renaming variables, generating derivative variables, applying transforms, etc.)”, wherein the BRI for a training data set generator engine encompasses the data pre-processor 910 that generates data for training by cleaning outlier data), a set of data samples of features of at least one (i.e. para. [0052], Fig. 2, “At operation 202, the system accesses fracturing data. In some examples, the fracturing data may be log data, for example, stored on and retrieved from a data store. In some examples, the fracturing data may be received from a live well such as a test well, laboratory well, producing well, or the like”, wherein the BRI for a generation device encompasses the well device with a measurable amount of pressure output or rate of volume generated by a well device), when providing at least a first (i.e. para. [0046], data channels for measurements of pressures, pump rates, pumped fluid volumes, and proppant concentrations, among other possible fields, may be recorded at, for example and without imputing limitation, one-second intervals); determining, for at least some of the data samples, a distance between the features of the data sample and features of other data samples (i.e. para. [0054], “determining a threshold value, such as a maximum deviation from a mean or the like, and removing data that is exceeds that threshold (e.g., removing data that is more than a standard deviation from a determined mean, etc.)”, wherein some data samples may have a distance of more than one standard deviation from a feature mean of the fractured data set)), wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device; identifying at least one outlier data sample of the data sample set based on the distances determined for at least some of the set of data samples (i.e. para. [0054], “Pre-processing includes cleaning the data, which may entail removing outlier values and the like”, wherein outliers are identified as data with a distance more than a standard deviation away from a determined mean and are removed); and generating, by the training data set generator engine, a training data set for a machine learning model, wherein the training data set includes the set of data samples excluding at least one of the at least one outlier data sample (i.e. para. [0055], “At operation 206, one or more fracturing stages are selected from the pre-processed data which may be used to generate a trained model for predicting”, wherein the BRI for a training data set encompasses how the pre-processed data has been cleaned to exclude outlier data and is to be used for training a predictive machine learning model); and
training (i.e. para. [0056], At operation 208, a training dataset, a validation datamaxset, and a testing dataset corresponding to the selected stages are extracted from the pre-processed data. As discussed above, in one example, 66% of the extracted data may be used for the training dataset), by a machine learning trainer using the training data set (i.e. para. [0080], “Model training process 918 includes a trainer 920 and a validator 922 which may iteratively train, validate, and, if necessary, retrain or further train neural network 924 as discussed above. In particular, trainer 920 may use a designated and appropriately pre-processed portion of the data to update model weights (e.g., via back propagation, equilibrium propagation, etc.) based on error values against labels of the training data), the machine learning model to predict a second power generation device (i.e. para. [0083], “Based on the filtered data, trained model 916 may generate event flag predictions (e.g., stage start/stop times, ISIP values, etc.)”, wherein the BRI for a prediction of a second output of the at least one generation device encompasses how the model has been trained to predict an instantaneous shut-in pressure (ISIP) value, wherein ISIP is commonly used to determine, among other measures, minimum principal stress in a downhole environment).
Lopez sets the stage for a training data generator engine that generates training data that is used by a machine learning trainer to train a model to predict an output value for a device that generates a real and measurable output. However, Lopez may not explicitly teach
At least one power generation device when producing at least a first power output;
wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device
to predict a second power output of the at least one power generation device.
However, Granade also teaches
At least one power generation device when producing at least a first power output (i.e. para. [0037], Referring to the framework depicted in FIG. 1, according to some embodiments, a network of meters 102 can each measure, record, and transmit raw data associated with an electrical grid);
determining, for at least some of the data samples, a distance between the features of the data sample and features of other data samples (i.e. para. [0039], “An Exception may refer to raw data indicative of values that are outside of a predetermined range. For example, a Voltage Exception may occur if a voltage level measured by a meter is outside of a predetermined range of acceptable levels.”, wherein the BRI for features of the data sample encompasses a voltage measurement of a power generation device is outside of a predetermined range of acceptable levels. Wherein the BRI for features of other data samples encompasses voltage measurements within of a predetermined range of acceptable levels. Wherein a distance may be found between voltages above an acceptable threshold compared to a threshold, and when the distance puts a voltage above the threshold, the measurement is found to be anomalous)),
Granade further teaches
wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode (i.e. para. [0037], “Voltage data may include a maximum voltage, a minimum voltage, and an average voltage that the meter measured over a predetermined measuring period”, wherein the BRI for a MPPG mode encompasses how a grid may be operating in a MPPG mode as it is capable of outputting and recording a maximum voltage) of the at least one power generation device (i.e. para. [0039], “Meter Malfunction may occur if a meter measures and/or reports energy consumption values that are outside of a predetermined range of acceptable values or if the energy usage consumption pattern exhibits an unusual pattern, such as one or more irregular spikes in consumption over time”, wherein it is noted that the BRI for a power output measurement during a MPPG is quite broad as it is unclear to what degree a maximum potential is measured as any power generation device with a safe acceptable output threshold could operating in a maximum potential power generation mode for safety. Thus the BRI for a power output measurement during a MPPG mode would encompass any measurement of a generator operating in a mode for safe maximum potently power generation); It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device, to the feature extraction and training for identifying and predicting outliers of Lopez, with how a identification of outliers may be based on a comparison to a maximum measured voltage during a potential mode in which voltage is measured over a period, as taught by Granade. One would have been motivated to combine Granade with Lopez as the combination increases the likelihood of identifying any anomalous data existing in the raw data, and accurately diagnose any issues associated with any anomalous data.
While Lopez-Granade teach power generation devices that produce measurable outputs during a MPPG mode and compares measured values that may be used to train a predictive model, Lopez Granade may not explicitly teach the
machine learning model to predict a second power output of the at least one power generation device
However, Maheswari also teaches
At least one power generation device when producing at least a first power output (i.e. para. [0056], The historical production capability data 304 may include information describing a previous energy production capability of the energy generating asset”, wherein the BRI for a first power output encompasses the historical energy production data for an asset);
Maheswari further teaches a
machine learning model to predict a second power output of the at least one power generation device (i.e. para. [0060], “the historical production capability data 304 associated with the particular energy generating asset. During the particular embodiment, an energy production value for each energy production value is predicted using the current meteorological data 202 and the trained model corresponding to the particular energy generating asset”, wherein the BRI to predict a second power output encompasses to generate a prediction for the asset).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add to predict a second power output of the at least one power generation device, with how data may be pre-processed and used as training data in preparation for use in machine learning prediction of an asset of Lopez-Granade, with how a historical measure of power output may be used to predict a second power output of the power generation device, as taught by Maheswari. One would have been motivated to combine Maheswari with Lopez-Granade as the combination provides operators with addition information that may be used to select from options such as energy storage, cryptocurrency mining, maintenance, or to maximize site revenue potential.
Claim 2:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
wherein the features of the data samples include at least one of a solar irradiance feature, a cloud coverage feature, an ambient temperature feature, a humidity feature, a geographic location feature, a power generation device type feature, a data sample time feature, and a power output feature (i.e. para. [0098], “FIG. 16 depicts an ISIP value heat map 1600 which may be generated based on the resulting ISIP values for multiple well stages. In effect, the ISIP values can be used to provide visualizations and interfaces for quickly determining well characteristics (e.g., pressure build ups, stresses, rock type barriers, etc.)”, wherein the BRI for a power output feature encompasses the PSI buildup due to well device characteristic features).
Claim 5:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
wherein the identifying is based at least in part on applying a rule to each of at least one of the data samples of the set (i.e. para. [0093], “limit filters of equations (1) and (2) are applied, outlying data points are removed (e.g., data outside of a modified standard deviation is removed)“, wherein the BRI for a rule encompasses applying a filter equation limiting the data to within a modified standard deviation of a mean).
Claim 6:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches wherein identifying the at least one outlier data sample includes ranking the data samples by the distances and identifying, as the outlier data samples, data samples that are within a top portion of the ranking (i.e. para. [0093], “FIGS. 14A-B respectively depict a plot 1400 of a dataset with outliers left in data for TR 1402A and SR 1404B and a plot 1450 of the same dataset with outliers filtered out from TR 1452A and SR 1452B, according to lower and upper limit filters according to equations (1) and (2)… when limit filters of equations (1) and (2) are applied, outlying data points are removed (e.g., data outside of a modified standard deviation is removed)”, wherein the BRI for ranking the data samples encompasses ranking the data samples into plots that are above a certain distance, such as further than a standard deviation from an average mean as determined by equations (1) and (2). Wherein the samples that are within a top portion of data points above the certain distance are filtered as outliers and are removed).
Claim 9:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
wherein identifying the at least one outlier data sample includes identifying the data samples having a distance that is above a threshold distance (i.e. para. [0093], “FIGS. 14A-B respectively depict a plot 1400 of a dataset with outliers left in data for TR 1402A and SR 1404B and a plot 1450 of the same dataset with outliers filtered out from TR 1452A and SR 1452B, according to lower and upper limit filters according to equations (1) and (2)… when limit filters of equations (1) and (2) are applied, outlying data points are removed (e.g., data outside of a modified standard deviation is removed)”, wherein the BRI for a threshold distance encompasses data that may be further than a standard deviation from an average mean as determined by equations (1) and (2). Wherein the samples that are above the threshold distance are filtered as outliers and are removed).
Claim 10:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Granade further teaches
wherein identifying the at least one outlier data sample is based on a comparison of a power output feature of the data sample and a power output measurement during the MPPG mode of the at least one power generation device associated with the data sample (i.e. para. [0037-0039], “to identify anomalous data… a network of meters 102 can each measure, record, and transmit … voltage data… Voltage data may include a maximum voltage… that measuring period… identification of Exceptions (i.e., anomalous data) existing in the raw data may made. An Exception may refer to raw data indicative of values that are outside of a predetermined range… a Voltage Exception may occur if a voltage level measured by a meter is outside of a predetermined range of acceptable levels”, wherein the BRI for a power output feature of the data sample encompasses a measured voltage that is outside a predetermined range of a maximum voltage and the BRI for a power output measurement during a maximum potential power generation mode encompasses a measured maximum voltage during a potential mode of a predetermined time period. Wherein it is noted that if a measured voltage is compared to a maximum possible voltage and found to be outside the range of the maximum voltage, the data indicating the measured voltage may be found to be an anomalous outlier) .
Claim 11:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
further comprising selecting, from the features, a subset of features for training the machine learning model (i.e. para. [0059], Fig. 2, At operation 212, features are selected from the training, validation, and testing datasets for a machine learning model to consume through at least a portion of the training process. As discussed above, in at least some examples, TP, SR, CV, and PC may be initially selected. In some examples, additional and/or alternative features may be selected throughout the training process).
Claim 13:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
further comprising retraining the machine learning model based on an update of the training data set (i.e. para. [0045], the model may benefit from retraining periodically with new field data to improve the prediction robustness and maintain high accuracy).
Claim 14:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
further comprising updating at least one hyperparameter associated with the machine learning model during a retraining of the machine learning model (i.e. para. [0056-0059, 0080], “in one example, 66% of the extracted data may be used for the training dataset, 8% may be used for the validation dataset, and 26% may be used for the testing dataset. While other distributions of extracted data to training dataset, validation dataset, and testing dataset (e.g., hyper-parameters) may be used, the ratio above was selected based on empirical determination of its efficacy in training certain machine learning models… At operation 212, features are selected from the training, validation, and testing datasets for a machine learning model to consume through at least a portion of the training process… In some examples, additional and/or alternative features may be selected throughout the training process … Model training process 918 includes a trainer 920 and a validator 922 which may iteratively train, validate, and, if necessary, retrain or further train neural network 924 as discussed above”, wherein it is noted that the BRI for at least one hyper-parameter encompasses the at least one feature selected from the testing dataset for the machine learning model to consume. It is noted that upon different retraining iterations in which the pre-processed fractured data is updated, the at least one feature selected from the testing dataset for the machine learning model to consume may also be updated during such a retraining).
Claim 15:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
further comprising predicting a power output of a first power generation device, the predicting being based on an output of the machine learning model in response to features of the first power generation device (i.e. para. [0069], “FIG. 6 depicts an example of hydraulic fracturing stage start time and stage end time predictions generated by a trained machine learning model such as discussed above… the trained machine learning model identifies an accurate start time, labeled as start time flag 608A, and an accurate end time, … The trained machine learning model processes features from TP 602, SR 604, and CV 606 pre-processed plot lines).
Claim 17:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 15.
Lopez further teaches
wherein the power output is predicted for the first power generation device during a maximum potential power generation mode of the first power generation device based on the features of the first power generation device (i.e. para. [0047], Fig. 6, “start time flag 101 and end time flag 110 may limit data included when determining average and/or maximum pressures, rates, and/or concentrations, as well as cumulative volumes and pumping time”, wherein the BRI for a maximum potential power generation mode encompasses a mode in which a maximum potential power is recorded during a time period. Wherein it is noted that Fig. 6 is a graph showing a predicted time window that displays maximum pressures, rates, and/or concentrations, as well as cumulative volumes and pumping time using the trained machine learning model process features).
Claim 19:
Lopez teaches a system, comprising: a memory that stores instructions, and a processor that is coupled to the memory and, when executing the instructions (i.e. para. [0101], Main memory 1708 may include one or more memory cards and a control circuit (not depicted), or other forms of removable memory, and may store various software applications including computer executable instructions, that when run on the processor 1704, implement the methods and systems set out herein), is configured to: receive, by a training data set generator engine (i.e. para. [0054], Pre-processing includes cleaning the data, which may entail removing outlier values and the like, and other modifications to the data to increase the efficiency of training (e.g., renaming variables, generating derivative variables, applying transforms, etc.)”, wherein the BRI for a training data set generator engine encompasses the data pre-processor 910 that generates data for training by cleaning outlier data), a set of data samples of features of at least one (i.e. para. [0052], Fig. 2, “At operation 202, the system accesses fracturing data. In some examples, the fracturing data may be log data, for example, stored on and retrieved from a data store. In some examples, the fracturing data may be received from a live well such as a test well, laboratory well, producing well, or the like”, wherein the BRI for a power generation device encompasses the well device with a measurable amount of pressure power or rate of volume generated by a well device), when providing at least a first (i.e. para. [0046], data channels for measurements of pressures, pump rates, pumped fluid volumes, and proppant concentrations, among other possible fields, may be recorded at, for example and without imputing limitation, one-second intervals),
determine, for each data sample, a distance between the features of the data sample and features of other data samples (i.e. para. [0054], “determining a threshold value, such as a maximum deviation from a mean or the like, and removing data that is exceeds that threshold (e.g., removing data that is more than a standard deviation from a determined mean, etc.)”, wherein some data samples may have a distance of more than one standard deviation from a feature mean of the fractured data set)), wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device,
identify at least one outlier data sample of the data sample set, based on the distance determined for each data sample (i.e. para. [0054], “Pre-processing includes cleaning the data, which may entail removing outlier values and the like”, wherein outliers are identified as data with a distance more than a standard deviation away from a determined mean and are removed),
generate, by the training data set generator engine, a training data set for a machine learning model, wherein the training data set includes the set of data samples excluding at least one of the at least one outlier data sample (i.e. para. [0055], “At operation 206, one or more fracturing stages are selected from the pre-processed data which may be used to generate a trained model for predicting”, wherein the BRI for a training data set encompasses how the pre-processed data has been cleaned to exclude outlier data and is to be used for training a predictive machine learning model), and
train, by a machine learning trainer using the training data set (i.e. para. [0080], “Model training process 918 includes a trainer 920 and a validator 922 which may iteratively train, validate, and, if necessary, retrain or further train neural network 924 as discussed above. In particular, trainer 920 may use a designated and appropriately pre-processed portion of the data to update model weights (e.g., via back propagation, equilibrium propagation, etc.) based on error values against labels of the training data), the machine learning model to predict a second (i.e. para. [0083], “Based on the filtered data, trained model 916 may generate event flag predictions (e.g., stage start/stop times, ISIP values, etc.)”, wherein the BRI for a prediction of a second output of the at least one generation device encompasses how the model has been trained to predict an instantaneous shut-in pressure (ISIP) value, wherein ISIP is commonly used to determine, among other measures, minimum principal stress in a downhole environment).
Lopez sets the stage for a training data generator engine that generates training data that is used by a machine learning trainer to train a model to predict an output value for a device that generates a real and measurable output. However, Lopez may not explicitly teach
At least one power generation device when producing at least a first power output;
wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device,
to predict a second power output of the at least one power generation device.
However, Granade also teaches
At least one power generation device when producing at least a first power output (i.e. para. [0037], Referring to the framework depicted in FIG. 1, according to some embodiments, a network of meters 102 can each measure, record, and transmit raw data associated with an electrical grid);
determining, for at least some of the data samples, a distance between the features of the data sample and features of other data samples (i.e. para. [0039], “An Exception may refer to raw data indicative of values that are outside of a predetermined range. For example, a Voltage Exception may occur if a voltage level measured by a meter is outside of a predetermined range of acceptable levels.”, wherein the BRI for features of the data sample encompasses a voltage measurement of a power generation device is outside of a predetermined range of acceptable levels. Wherein the BRI for features of other data samples encompasses voltage measurements within of a predetermined range of acceptable levels. Wherein a distance may be found between voltages above an acceptable threshold compared to a threshold, and when the distance puts a voltage above the threshold, the measurement is found to be anomalous).
Granade further teaches
wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode (i.e. para. [0037], “Voltage data may include a maximum voltage, a minimum voltage, and an average voltage that the meter measured over a predetermined measuring period”, wherein the BRI for a MPPG mode encompasses how a grid may be operating in a MPPG mode as it is capable of outputting and recording a maximum voltage) of the at least one power generation device (i.e. para. [0039], “Meter Malfunction may occur if a meter measures and/or reports energy consumption values that are outside of a predetermined range of acceptable values or if the energy usage consumption pattern exhibits an unusual pattern, such as one or more irregular spikes in consumption over time”, wherein it is noted that the BRI for a power output measurement during a MPPG is quite broad as it is unclear to what degree a maximum potential is measured as any power generation device with a safe acceptable output threshold could operating in a maximum potential power generation mode for safety. Thus the BRI for a power output measurement during a MPPG mode would encompass any measurement of a generator operating in a mode for safe maximum potently power generation); It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode of the at least one power generation device, to the feature extraction and training for identifying and predicting outliers of Lopez, with how a identification of outliers may be based on a comparison to a maximum measured voltage during a potential mode in which voltage is measured over a period, as taught by Granade. One would have been motivated to combine Granade with Lopez as the combination increases the likelihood of identifying any anomalous data existing in the raw data, and accurately diagnose any issues associated with any anomalous data.
While Lopez-Granade teach power generation devices that produce measurable outputs during a MPPG mode and compares measured values that may be used to train a predictive model, Lopez Granade may not explicitly teach the
machine learning model to predict a second power output of the at least one power generation device.
However, Maheswari also teaches
At least one power generation device when producing at least a first power output (i.e. para. [0056], The historical production capability data 304 may include information describing a previous energy production capability of the energy generating asset”, wherein the BRI for a first power output encompasses the historical energy production data for an asset);
Maheswari further teaches a
machine learning model to predict a second power output of the at least one power generation device (i.e. para. [0060], “the historical production capability data 304 associated with the particular energy generating asset. During the particular embodiment, an energy production value for each energy production value is predicted using the current meteorological data 202 and the trained model corresponding to the particular energy generating asset”, wherein the BRI to predict a second power output encompasses to generate a prediction for the asset).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to At least one power generation device when producing at least a first power output; to predict a second power output of the at least one power generation device, with how data may be pre-processed and used as training data in preparation for use in machine learning prediction of an asset of Lopez-Granade, with how a historical measure of power output may be used to predict a second power output of the power generation device, as taught by Maheswari. One would have been motivated to combine Maheswari with Lopez-Granade as the combination provides operators with addition information that may be used to select from options such as energy storage, cryptocurrency mining, maintenance, or curtailment to maximize site revenue potential.
Claim 22:
Claim 22 is the system claim reciting similar limitations to claim 15 and is rejected for similar reasons.
Claim 23:
Claim 23 is the system claim reciting similar limitations to Claim 1 and is rejected for similar reasons.
Claim 26:
Claim 26 is the system claim reciting similar limitations to Claim 15 and is rejected for similar reasons.
Claim 27:
Lopez teaches a computer-implemented method, comprising: receiving a set of data samples of features of a power generation device (i.e. para. [0052], Fig. 2, “At operation 202, the system accesses fracturing data. In some examples, the fracturing data may be log data, for example, stored on and retrieved from a data store. In some examples, the fracturing data may be received from a live well such as a test well, laboratory well, producing well, or the like”, wherein the BRI for a power generation device encompasses the well device with a measurable amount of pressure power or rate of volume generated by a well device) when providing at least a first (i.e. para. [0046], data channels for measurements of pressures, pump rates, pumped fluid volumes, and proppant concentrations, among other possible fields, may be recorded at, for example and without imputing limitation, one-second intervals); and processing the set of data samples using a trained machine learning model to predict a second (i.e. para. [0055], “At operation 206, one or more fracturing stages are selected from the pre-processed data which may be used to generate a trained model for predicting”, wherein the BRI to predict a power output encompasses how the predictive machine learning model generates Fig. 6, which is a predictive graph showing a time window that displays maximum pressures, rates, and/or concentrations, as well as cumulative volumes and pumping time using the trained machine learning model process features), wherein the trained machine learning model has been trained on a set of data samples excluding at least one outlier data sample i.e. para. [0054], “Pre-processing includes cleaning the data, which may entail removing outlier values and the like”, wherein outliers are identified as data with a distance more than a standard deviation away from a determined mean and are removed), and wherein the at least one outlier data sample has been determined based on a distance between features of the outlier data sample and features of other data samples of the set of data samples (i.e. para. [0054], “determining a threshold value, such as a maximum deviation from a mean or the like, and removing data that is exceeds that threshold (e.g., removing data that is more than a standard deviation from a determined mean, etc.)”, wherein some data samples may have a distance of more than one standard deviation from a feature mean of the fractured data set)).
Granade further teaches
wherein the features of other data samples include a power output measurement during a maximum potential power generation (MPPG) mode (i.e. para. [0037], “Voltage data may include a maximum voltage, a minimum voltage, and an average voltage that the meter measured over a predetermined measuring period”, wherein the BRI for a MPPG mode encompasses how a grid may be operating in a MPPG mode as it is capable of outputting and recording a maximum voltage) of the at least one power generation device (i.e. para. [0039], “Meter Malfunction may occur if a meter measures and/or reports energy consumption values that are outside of a predetermined range of acceptable values or if the energy usage consumption pattern exhibits an unusual pattern, such as one or more irregular spikes in consumption over time”, wherein it is noted that the BRI for a power output measurement during a MPPG is quite broad as it is unclear to what degree a maximum potential is measured as any power generation device with a safe acceptable output threshold could operating in a maximum potential power generation mode for safety. Thus the BRI for a power output measurement during a MPPG mode would encompass any measurement of a generator operating in a mode for safe maximum potently power generation);
Claim(s) 3 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20200065677 “Lopez”, further light of U.S. Patent Application Publication NO. 20190025365 “Granade”, and further in light of U.S. Patent Application Publication NO. 20230214703 “Maheswari”, as applied to claim 1 above, and further light of U.S. Patent Application Publication NO. 20210158949 “Chen”.
Claim 3:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez may not explicitly teach further comprising
normalizing the features of at least some of the data samples.
However, Chen teaches
normalizing the features of at least some of the data samples (i.e. para. [0046-0047], the data platform module 101a may also have data values which are normalized to remove any outliers, based on a set of features of the raw data... The system 101 also includes the data modeling module 101b, which is configured to perform one or more operations related to data preparation for use in machine learning (ML). These operations may include such as data labeling, ML pair generation, model training, transfer learning and the like).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add normalizing the features of at least some of the data samples, to the data cleaning methods of Lopez-Granade-Maheswari, with how a cleaned data may further be normalized in preparation for use in machine learning, as taught by Chen. One would have been motivated to combine Chen with Lopez-Granade-Maheswari as the combination provides results in an improvement in model performance by ensuring consistency and avoiding skewing.
Claim(s) 4, 7-8, 20, 24 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20200065677 “Lopez”, further light of U.S. Patent Application Publication NO. 20190025365 “Granade”, and further in light of U.S. Patent Application Publication NO. 20230214703 “Maheswari”, as applied to claims 1 and 19 above, and further light of U.S. Patent Application Publication NO. 20210158081 “Devitt”.
Claim 4:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez may not explicitly teach
wherein the identifying is based on a K-nearest-neighbor determination between the features of a first data sample and the features of other data samples.
However Devitt teaches,
wherein the identifying is based on a K-nearest-neighbor determination between the features of a first data sample and the features of other data samples (i.e. para. [0030], “variants of the technology can be tuned for a variety of different applications… the size and/or number of nearest neighbors that are searched and/or used for interpolation (e.g., 1.sup.st nearest neighbors, 2.sup.nd nearest neighbors, 3.sup.rd nearest neighbors, etc)”, wherein the BRI for K-nearest neighbor encompasses the comparing features of a first data sample with features of other data samples to find a vector representing a type of nearest neighbor).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add wherein the identifying is based on a K-nearest-neighbor determination between the features of a first data sample and the features of other data samples, to the distance calculating methods for determining an outlier of Lopez-Granade-Maheswari, with how a nearest neighbor distance metric calculation may be used for machine learning data, as taught by Devitt. One would have been motivated to combine Devitt with Lopez-Granade-Maheswari as the combination provides users with more versatile distance calculation metrics for applications in machine learning.
Claim 7:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
Lopez further teaches
wherein the distance determined for each data sample is based on a (i.e. para. [0054], “). Removing outliers, for example, involves determining a threshold value, such as a maximum deviation from a mean or the like, and removing data that is exceeds that threshold (e.g., removing data that is more than a standard deviation from a determined mean, etc.)”, wherein the BRI for the distance encompasses the distance between data sample and features of other data samples that is more than a standard deviation).
While Lopez teaches calculating a distance, Lopez may not explicitly teach a
Minkowski distance.
However, Devitt teaches
A Minkowski distance (i.e. para. [0056], The cost module preferably determines (e.g., calculates, selects, etc.) a cost value of a cost metric (e.g., distance metric). Examples of cost metrics that can be used include: the … Minkowski distance…The distance metric can be predetermined, learned from data (for example using a neural network or other machine learning algorithm), and/or otherwise selected).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add a Minkowski distance, to the distance calculating methods for determining an outlier of Lopez-Granade-Maheswari, with how a Minkowski distance calculation may be used to find a distance metric for machine learning data, as taught by Devitt. One would have been motivated to combine Devitt with Lopez-Granade-Maheswari as the combination provides users with more versatile distance calculation metrics for applications in machine learning.
Claim 8:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 1.
While Lopez teaches determining a distance, Lopez may not explicitly teach a
wherein the distance determined for each data sample is based on an arithmetic median of the distance between the features of the data sample and the features of other data samples.
However, Devitt teaches
wherein the distance determined for each data sample is based on an arithmetic median of the distance between the features of the data sample and the features of other data samples (i.e. para. [0137], Post-processing the correspondence map S700 can function to clean up the correspondence map results (e.g., remove digital artifacts), remove outliers (e.g., with an N×N median filter, such as a 3×3 median filter).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add wherein the distance determined for each data sample is based on an arithmetic median of the distance between the features of the data sample and the features of other data samples, to the distance calculating methods for determining an outlier of Lopez-Granade-Maheswari, with how a removal of outliers may be based on a median distance filter, as taught by Devitt. One would have been motivated to combine Devitt with Lopez-Granade-Maheswari as the combination provides users with more versatile distance calculation metrics for applications in machine learning.
Claim 20:
Lopez, Granade, and Maheswari teach the system of claim 19.
Lopez may not explicitly teach
wherein the identifying is based on a K-nearest-neighbor determination between the features of each data sample and the features of the other data samples.
However Devitt teaches,
wherein the identifying is based on a K-nearest-neighbor determination between the features of each data sample and the features of the other data samples (i.e. para. [0030], “variants of the technology can be tuned for a variety of different applications… the size and/or number of nearest neighbors that are searched and/or used for interpolation (e.g., 1.sup.st nearest neighbors, 2.sup.nd nearest neighbors, 3.sup.rd nearest neighbors, etc)”, wherein the BRI for K-nearest neighbor encompasses the comparing features of a first data sample with features of other data samples to find a vector representing a type of nearest neighbor).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add wherein the identifying is based on a K-nearest-neighbor determination between the features of each data sample and the features of the other data samples, to the distance calculating methods for determining an outlier of Lopez-Granade-Maheswari, with how a nearest neighbor distance metric calculation may be used for machine learning data, as taught by Devitt. One would have been motivated to combine Devitt with Lopez-Granade-Maheswari as the combination provides users with more versatile distance calculation metrics for applications in machine learning.
Claim 24:
Claim 24 is the system claim reciting similar limitations to Claim 20 and is rejected for similar reasons.
Claim(s) 16 & 18 is/are rejected under 35 U.S.C. 103 as being unpatentable over U U.S. Patent Application Publication NO. 20200065677 “Lopez”, further light of U.S. Patent Application Publication NO. 20190025365 “Granade”, and further in light of U.S. Patent Application Publication NO. 20230214703 “Maheswari”, as applied to Claim 1 above, and further light of U.S. Patent Application Publication NO. 20120038329 “Hampo”.
Claim 16:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 15.
Lopez may not explicitly teach
initiating an action based on a difference between the power output predicted for the first power generation device and a power output measurement of the first power generation device.
However, Hampo teaches further comprising
initiating an action based on a difference between the power output predicted for the first power generation device and a power output measurement of the first power generation device (i.e. para. [0042-0043], Fig. 4, “The method (80) may further comprise measuring (92) a control section voltage, and performing (94) a diagnostic function based on a comparison of the predicted output voltage and the measured control section voltage”, wherein the BRI for initiating an action encompasses initiating a diagnostic function based on a difference between the predicted and measured power output of the device which may be controlling the output voltage of a power supply, which may further comprise determining (100) a temperature adjustment factor for adjusting the measured control section temperature).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add initiating an action based on a difference between the power output predicted for the first power generation device and a power output measurement of the first power generation device, to the review of predictive data of Lopez-Granade-Maheswari, with a diagnostic action may be initiated when a predictive value is different than a measured value for a power generating device, as taught by Hampo. One would have been motivated to combine Hampo with Lopez-Granade-Maheswari as the combination provides users with a more streamlined diagnostic process for fixing monitored devices.
Claim 18:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 15.
Lopez teaches further comprising operating one or both of a second power generation device or a power load device, wherein the operating is based on a predicted power output of the first power generation (i.e. para. [0083], “Based on the filtered data, trained model 916 may generate event flag predictions (e.g., stage start/stop times, ISIP values, etc.), which may be provided to downstream processes 924. In some examples, downstream processes 924 may include a graphical user interface (GUI) or the like for direct user access… downstream processes 924 include automated or third party systems accessed directly or via application programming interface (API) to, for example and without imputing limitation, determine drill controls, generate other predictions, trigger alerts, etc. ”, wherein the BRI for operating encompasses how a user may modify the operation of the well device using a GUI by undertaking downstream processes based on the displayed predictive data flags)
However, Hampo teaches further comprising
operating one or both of a second power generation device or a power load device, wherein the operating is based on a predicted power output of the first power generation device (i.e. para. [0041], “The method (80) may further comprise providing (86) a control section voltage based on the adjusted voltage command in order to supply an output voltage at the power supply output point that is substantially equal to the desired output voltage”, wherein the power device may be operated at an adjusted voltage based on a predicted output voltage).
Claim(s) 28 is/are rejected under 35 U.S.C. 103 as being unpatentable over U.S. Patent Application Publication NO. 20200065677 “Lopez”, further light of U.S. Patent Application Publication NO. 20190025365 “Granade”, and further in light of U.S. Patent Application Publication NO. 20230214703 “Maheswari”, as applied to claim 27 above, and further light of U.S. Patent Application Publication NO. 20130197834 “Maki”.
Claim 28:
Lopez, Granade, and Maheswari teach the computer-implemented method of claim 27.
While Lopez teaches identifying the at least one outlier data sample, Granade teaches whether the power generation device is operating in a maximum potential power generation mode, and Maheswari teaches the power output predicted by the trained machine learning model and a measured power output of the power generation device (i.e. para. [0047], FIG. 2 also includes predicting 206 an energy production value produced by each energy generating asset at a predetermined time. In the example method depicted in FIG. 2, the energy production value produced by each energy generating asset), Lopez-Granade-Maheswari and may not explicitly teach
further comprising determining, based on the second predicted power output predicted by the trained machine learning model and a measured power output of the power generation device, whether the power generation device is operating in a maximum potential power generation mode.
However, Maki teaches
further comprising determining, based on the second predicted power output (i.e. para. [0023-0024], the estimated maximum power point voltage is compared with the determined maximum power point. Based on the comparison between the estimated and determined maximum power points… When the estimated and determined maximum power point voltages differ more than a pre-determined amount, then it is concluded that there are two maximum power points), whether the power generation device is operating in a maximum potential power generation mode (i.e. para. [0023], “it is determined if the PV power generator is operating under uniform conditions or in conditions with two maximum power points”, wherein the BRI for a maximum potential power generation mode encompasses how it may be determined that a power generator is operating under conditions that product a maximum power readings as opposed to uniform conditions that would not produce maximum power point readings).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to add further comprising determining, based on the predicted power output and a measured power output of the power generation device, whether the power generation device is operating in a maximum potential power generation mode, to the power analysis and machine learning prediction of Lopez-Granade-Maheswari, with how the determination that a device is in a mode that results in maximal power output being generated may be identified by comparing an estimated power output to a measured power output, as taught by Maki. One would have been motivated to combine Maki with Lopez-Granade-Maheswari as the combination saves user’s time by automating a scanning of device power settings and more quickly lets user’s ascertain a status of a device.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
U.S. Patent Application Publication No. 20040264083 “Mansingh”, teaches in para. [0222], the In certain exemplary embodiments of VSTLP module 1600, six separate neural networks can be used to cover the 24-hour day. Multiple neural networks can be utilized when there are different load dynamics during different periods within the same day. Making a neural network responsible for a smaller portion of the day can increase overall prediction accuracy. In certain exemplary embodiments, the neural networks can be trained from historical day load patterns. The day patterns that are to be used can come from the weather-adaptive load forecast and/or from manual entry.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to DAVID H TAN whose telephone number is (571)272-7433. The examiner can normally be reached M-F 7:30-4:30.
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, Cesar Paula can be reached at (571) 272-4128. 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.
/D.T./Examiner, Art Unit 2145
/CESAR B PAULA/Supervisory Patent Examiner, Art Unit 2145
Prosecution Insights