Condition-based asset management

Abstract

Methods, systems, and apparatus for predicting a degree of success of a maintenance cycle performed on an asset based on a plurality of operating parameters. A predictive model may be trained and tested based on the plurality of operating parameters. The prediction model may be configured to output a prediction indicative of the degree of success of the maintenance cycle.

Description

[0001] Cross-reference to related applications

[0002] This application claims priority to U.S. Non-provisional Application No. 18 / 149,725, filed January 4, 2023, which is incorporated herein by reference in its entirety. Background Art

[0003] An electric power system comprises a network of electrical components or power system equipment configured to supply, transmit, and / or use electricity. For example, an electric power grid (also known as a power distribution network) includes generators, a transmission system, and / or a distribution system. Generators or power plants are configured to generate electricity from combustible fuels (e.g., coal, natural gas, etc.) and / or non-combustible fuels (e.g., wind, solar, nuclear, etc.). The transmission system is configured to carry or transmit the electricity from the generators to the loads. The distribution system is configured to feed the supplied electricity to nearby homes, businesses, and / or other institutions. In addition to other electrical components, such an electric power system may also include one or more transformers configured to convert or transform electricity at one voltage (e.g., the voltage used to transmit the electricity) into electricity at another voltage (e.g., the voltage required by the loads receiving the electricity). Depending on the size of the power system and / or the loads applied to the transformers, the cost of purchasing a transformer can range from a few thousand dollars to over a million dollars.

[0004] Utilities can therefore benefit greatly from using machine learning methods and models to determine the degree of success associated with maintenance cycles, particularly with respect to transformers. However, one of the biggest challenges facing the use of machine learning is the lack of availability of large, annotated datasets. Annotation of data is not only expensive and time-consuming, but also highly dependent on the availability of expert observers. The limited amount of training data can inhibit the performance of supervised machine learning algorithms, which typically require large amounts of training data to avoid overfitting. To date, efforts have been made to extract as much information as possible from the available data. One area where large, annotated datasets are particularly lacking is the analysis of operational data associated with unit transformers. The ability to analyze transformer operational data to predict the degree of success of maintenance cycles is critical to managing the maintenance cycles of these transformers. However, in many cases, there is not enough data available to train machine learning algorithms to accurately predict the degree of success of maintenance cycles. Summary of the Invention

[0005] It is to be understood that both the following general description and the following detailed description are exemplary and explanatory only and are not restrictive.

[0006] In an embodiment, a method is disclosed, comprising: determining operational data associated with a plurality of operational parameters associated with an asset, wherein the plurality of operational parameters comprises one or more groups of operational parameters, and wherein each group of operational parameters is labeled according to a feature score; determining a plurality of feature scores for a predictive model based on the operational data; training the predictive model based on a first portion of the operational data according to the plurality of feature scores; testing the predictive model based on a second portion of the operational data; and outputting the predictive model based on the testing.

[0007] In an embodiment, a method is disclosed, comprising: determining a time series of data associated with the asset, wherein the time series comprises one or more time periods; performing analysis on each time period of the data in the one or more time periods of the data; and generating the operational data based on the analysis of each time period of the data, wherein the operational data comprises a data set associated with each time period.

[0008] In an embodiment, a method is disclosed, the method comprising: determining one or more operational data sets including at least one operational parameter of the plurality of operational parameters based on the plurality of operational parameters; and generating the operational data based on the one or more operational data sets.

[0009] In an embodiment, a method is disclosed, comprising: determining a baseline characteristic score for each group of operating parameters in the plurality of operating parameters; labeling the baseline characteristic score for each group of operating parameters in the plurality of operating parameters as the characteristic score associated with each group of operating parameters; and generating the operational data based on the labeled baseline characteristic score.

[0010] In an embodiment, a method is disclosed, comprising: receiving operating parameter data associated with multiple operating parameters of an asset, wherein the multiple operating parameters are determined during operation of the asset; providing the operating parameter data to a predictive model; and determining a predictive score associated with a maintenance cycle performed on the asset based on the predictive model.

[0011] Additional advantages will be set forth in part in the description which follows, or may be learned by practice.These advantages will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings, which are incorporated in and constitute a part of this specification, serve to explain the principles of the methods and systems described herein:

[0013] Figure 1 a flowchart illustrating an example method;

[0014] Figure 2 An example machine learning system is shown;

[0015] Figure 3 A flowchart illustrating an example machine learning method;

[0016] Figure 4 A block diagram illustrating an example computing device;

[0017] Figure 5 a flowchart illustrating an example method; and

[0018] Figure 6 A flow chart illustrating an example method is shown. DETAILED DESCRIPTION

[0019] As used in the specification and the appended claims, the singular forms "a," "an," and "the" include plural referents unless the context clearly dictates otherwise. Ranges may be expressed herein as from "about" one particular value and / or to "about" another particular value. When such ranges are expressed, another configuration includes from the one particular value and / or to the other particular value. Similarly, when values ​​are expressed as approximations by use of the antecedent "about," it should be understood that the particular value forms another configuration. It should also be understood that the endpoints of each range are significant with respect to the other endpoint, as well as independently of the other endpoint.

[0020] “Optional” or “optionally” means that the subsequently described event or circumstance may or may not occur, and that the description includes instances where said event or circumstance occurs and instances where it does not.

[0021] Throughout the detailed description and claims of this specification, the word "comprise" and variations thereof, such as "comprising" and "comprises," mean "including, but not limited to," and are not intended to exclude, for example, other components, integers, or steps. "Exemplary" means "an example of" and is not intended to convey an indication of a preferred or ideal configuration. "For example" is not limiting but is for explanatory purposes.

[0022] It should be understood that when describing combinations, subsets, interactions, groups, etc. of components, while specific reference may not be made to explicitly describing each different individual and collective combination and arrangement of these components, each is specifically contemplated and described herein. This applies to all portions of this application, including but not limited to the steps in the described methods. Therefore, if there are various additional steps that can be performed, it should be understood that each of these additional steps can be performed using any specific configuration or combination of configurations of the described methods.

[0023] As will be appreciated by those skilled in the art, hardware, software, or a combination of software and hardware may be implemented. Furthermore, the methods and systems may take the form of a computer program product on a computer-readable storage medium (e.g., non-transitory) having processor-executable instructions (e.g., computer software) embodied in the storage medium. Any suitable computer-readable storage medium may be utilized, including a hard disk, CD-ROM, optical storage device, magnetic storage device, memristor, non-volatile random access memory (NVRAM), flash memory, or a combination thereof.

[0024] Throughout this application, reference is made to block diagrams and flow charts. It should be understood that each block of the block diagrams and flow charts, and combinations of blocks in the block diagrams and flow charts, respectively, can be implemented by processor-executable instructions. These processor-executable instructions can be loaded onto a general-purpose computer, a special-purpose computer, or other programmable data processing device to produce a machine, such that the processor-executable instructions executed on the computer or other programmable data processing device produce means for implementing the functions specified in one or more flow chart blocks.

[0025] These processor-executable instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to function in a specific manner such that the processor-executable instructions stored in the computer-readable memory produce an article of manufacture containing processor-executable instructions for implementing the functions specified in one or more flowchart blocks. The processor-executable instructions may also be loaded onto a computer or other programmable data processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process such that the processor-executable instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one or more flowchart blocks.

[0026] The blocks of the block diagrams and flow charts support combinations of devices for performing the specified functions, combinations of steps for performing the specified functions, and program instruction means for performing the specified functions. It should also be understood that each block of the block diagrams and flow charts, and combinations of blocks in the block diagrams and flow charts, can be implemented by a computer system based on dedicated hardware or a combination of dedicated hardware and computer instructions that performs the specified functions or steps.

[0027] Methods and systems are described for generating a machine learning classifier for predicting the success of a maintenance cycle performed on an asset (e.g., a transformer) based on multiple operating parameters associated with the asset. Machine learning (ML) is a subfield of computer science that enables computers to learn without being explicitly programmed. Machine learning platforms include, but are not limited to, naive Bayes classifiers, support vector machines, decision trees, neural networks, and the like. In an example, operational data can be generated based on multiple operating parameters of the asset. The multiple operating parameters can include one or more groups of operating parameters. In an example, feature scores can be generated for the one or more groups of operating parameters. As another example, the feature scores can be analyzed to determine a predictive score that indicates the success of a maintenance cycle performed on the asset. The feature scores can include a metric that indicates the success of a maintenance cycle performed on the asset based on the one or more groups of operating parameters.

[0028] Figure 1 A flow chart is shown of an example method 100 for generating a predictive model, the method including determining operational data associated with a plurality of operational parameters of an asset (e.g., a transformer) at step 110, determining a plurality of feature scores for a predictive model based on the operational data at step 120, and generating the predictive model based on the plurality of feature scores at step 130.

[0029] The operational data may include one or more data sets based on multiple operational parameters of the asset. Each of the one or more data sets may include data indicating a time series of data associated with the multiple parameters of the asset. The multiple operational parameters may include one or more of the following: a power parameter, a voltage parameter, a current parameter, a capacity parameter, a heat parameter, a cooling pipe parameter, a fuel tank parameter, a sunlight duration parameter, a transformer height, a manufacturing date, manufacturer data, an installation date, vehicle traffic density, or an air temperature parameter. The multiple operational parameters may include one or more groups of operational parameters. Each group of operational parameters in the multiple operational parameters may be labeled according to a characteristic score. The characteristic score may include a metric indicating the degree of success of a maintenance cycle performed on the asset based on the one or more groups of operational parameters. For example, the metric may include a value (e.g., 0-10) indicating the level of success associated with a maintenance cycle performed on the asset. This value may be compared to a threshold (e.g., 1-10) indicating whether the maintenance cycle was successful. For example, if the metric is above the threshold, the maintenance cycle may be determined to be unsuccessful. If the metric is below the threshold, the maintenance cycle may be determined to be successful. As an example, if the maintenance cycle is determined to be successful, the metric may include a value of 1, and if the maintenance cycle is determined to be unsuccessful, the metric may include a value of 0.

[0030] Determining operational data associated with a plurality of operational parameters at step 110 may include downloading / obtaining / receiving one or more operational parameter data sets obtained from various sources, including recent publications and / or publicly available databases. As an example, the operational data may be determined based on determining a time series of data associated with a plurality of operational parameters of the asset, wherein the time series may include one or more time periods. Analysis may be performed on each time period of the data, wherein the data may be transformed based on the analysis of each time period of the data and provided as input data for determining / generating a predictive model. The operational data may be transformed to address any imbalances in the model data. For example, the operational data may be transformed based on one or more methods, such as an interpolation method, a method for handling outliers, a grouping method, a logarithmic transformation method, a data aggregation method, or a scaling method. In an example, the transformed operational data may increase the data resolution by a factor of 15.

[0031] As another example, the operational data can be transformed or concatenated with digital representations of corresponding computational variables and transformed or concatenated into a single concatenated digital representation (e.g., a concatenated vector). The concatenated vector can describe the operational data and the corresponding computational variables as a single digital vector, fingerprint, representation, etc. The concatenated vector can be passed to one or more machine learning-based models.

[0032] As another example, operational data can be determined (e.g., generated) based on one or more operational data sets, where the one or more operational data sets can include at least one operational parameter from a plurality of operational parameters. As another example, operational data can be determined based on a baseline characteristic score associated with each set of operational parameters. The baseline characteristic score can be labeled as the characteristic score associated with each set of operational parameters. The methods described herein can utilize one or more operational data sets to improve the identification of whether a maintenance cycle for an asset was successful.

[0033] about Figure 2 and Figure 3 Determining a plurality of feature scores for a prediction model based on operational data at step 120 and generating a prediction model based on the plurality of feature scores at step 130 are described.

[0034] For example, a predictive model (e.g., a machine learning classifier) ​​can be generated to determine a predictive score that indicates the success of a maintenance cycle performed on an asset (e.g., a transformer). The predictive model can be trained based on operational data (e.g., one or more operational datasets). The operational datasets can include time series datasets associated with operational parameters of the asset, such as power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, tank parameters, sunlight duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density, or air temperature parameters. The operational data can be associated with a baseline characteristic score associated with one or more sets of operational parameters, wherein the baseline characteristic score can indicate the success of a maintenance cycle performed on the asset. The baseline characteristic score can be associated with research from various sources across multiple platforms involving the utility, such as work order preventive / corrective maintenance, transformer oil dissolved gas analysis (DGA), tap controller oil DGA, inspection data, or equipment nameplate data at varying temporal resolutions. In an example, the baseline characteristic score can be associated with historical data, including historical sensor data derived from one or more sensors communicating with components of the asset and / or historical field test data derived from one or more field tests performed on components of the asset. In an example, one or more prediction scores for a prediction model may be generated based on the operational data.

[0035] The historical data may also relate to only a subset of the components of the asset (e.g., transformers). For example, the historical data may relate to a particular class or type of transformer (e.g., configured to convert voltage between a first voltage or first voltage range and a second voltage or second voltage range).

[0036] Figure 2An example system 200 is shown, configured to use machine learning techniques to train at least one machine learning-based classifier 234 based on analysis of one or more transformed / preprocessed training datasets 210 / 220 by a machine learning module 230. The at least one machine learning-based classifier is configured to classify baseline feature data into a feature score or metric associated with the degree of success of a maintenance cycle performed on an asset (e.g., a transformer). For example, the machine learning module 230 may include an automated machine learning module 230. The automated machine learning module 230 may automatically perform feature selection 231, model selection 232, and parameter selection 233 without human interaction. The training dataset 210 may include operational data, wherein the operational data may include one or more operational datasets associated with one or more sets of operational parameters and / or labeled baseline feature scores. The one or more operational datasets may include one or more time series datasets associated with one or more sets of operational parameters. The one or more operational datasets 210 may be transformed 220 based on one or more methods to address any imbalances in the model / training data. For example, the operational data can be transformed using methods such as interpolation methods, methods for handling outliers, grouping methods, logarithmic transformation methods, data aggregation methods, or scaling methods. The transformed operational data can increase the data resolution by a factor of 15. The operational parameters can include one or more of the following: power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, fuel tank parameters, sunlight duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density, or air temperature parameters. In an example, the training dataset 210 can include labeled baseline feature scores associated with one or more operational datasets. The feature score can indicate a metric associated with a maintenance cycle performed on an asset. For example, the metric can include a value (e.g., 0-10) indicating a level of success associated with a maintenance cycle performed on an asset. The value can be compared to a threshold (e.g., 1-10) indicating whether the maintenance cycle was successful. For example, if the metric is above the threshold, then the maintenance cycle can be determined to be unsuccessful. If the metric is below the threshold, then the maintenance cycle can be determined to be successful.

[0037] The automated machine learning module 230 can train a machine learning-based classifier 234 by extracting a set of features from the operational data (e.g., one or more operational data sets) in the transformed training data sets 210 / 220 according to one or more feature selection techniques. In addition, the automated machine learning module 230 can select a machine learning model 232. For example, the automated machine learning module 230 can select an appropriate machine learning model 232 based on a specific maintenance cycle to be performed on the asset and an evaluation to determine a metric indicating a level of success associated with the maintenance cycle. Furthermore, the automated machine learning module 230 can select appropriate parameters 233 based on the maintenance cycle being evaluated. This enhances the ability of the automated machine learning module 230 to appropriately tailor the trained classifier 234 based on the specific maintenance cycle evaluated and scored based on the trained classifier 234 (e.g., a metric indicating a level of success for the maintenance cycle).

[0038] The automated machine learning module 230 can extract feature sets from the training dataset 220 in various ways. The automated machine learning module 230 can perform feature extraction multiple times using different feature extraction techniques. In an example, the feature sets generated using the different techniques can each be used to generate a different machine learning-based classification model 234. In an example, the feature set with the highest quality metric can be selected for training. The automated machine learning module 230 can use the feature sets to build one or more machine learning-based classification models 234A-234N, wherein the one or more machine learning-based classification models are configured to indicate whether new data is associated with the operational evaluation of the transformer.

[0039] The training data set 220 can be analyzed to determine one or more features associated with one or more sets of operating parameters, wherein the one or more features can be further associated with one or more feature scores associated with the success of a maintenance cycle for an asset (e.g., a transformer). The one or more features and one or more feature scores associated with one or more sets of operating parameters can be considered features (or variables) in the context of machine learning. As used herein, the term "feature" can refer to any characteristic of a set or range of operating parameters that can be used to determine whether the set of operating parameters is associated with a feature score or a range of feature scores. For example, the features described herein can be associated with one or more sets of operating parameters.

[0040] The feature selection technique may include one or more feature selection rules. The one or more feature selection rules may include parameter occurrence rules. The parameter occurrence rules may include determining which operating parameter or group of operating parameters in the training data set 220 appears more than a threshold number of times, and identifying those parameters that meet the threshold as candidate features. For example, any parameter or group of parameters that appears greater than or equal to 50 times in the training data set 220 may be considered a candidate feature. Any parameter or group of parameters that appears less than 50 times may be excluded from consideration as a feature.

[0041] The one or more feature selection rules may include a significance rule. The significance rule may include determining based on baseline feature level data in the training dataset 220 (e.g., one or more operational datasets). The operational dataset may include data associated with the evaluation or analysis of one or more of the following: power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, fuel tank parameters, sunlight duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density, or air temperature parameters. When the baseline feature level data in the training dataset 220 is labeled based on one or more feature scores associated with one or more sets of operational parameters, the labels may be used to determine a feature score associated with a degree of success of a maintenance cycle performed on an asset based on the one or more sets of operational parameters.

[0042] A single feature selection rule may be applied to select features, or multiple feature selection rules may be applied to select features. In an example, feature selection rules may be applied in a cascaded manner, where the feature selection rules are applied in a specific order and are applied to the results of previous rules. For example, a parameter occurrence rule may be applied to the training dataset 220 to generate a first feature list. A significance rule may be applied to features in the first feature list to determine which features of the first list satisfy the significance rule in the training dataset 220 and generate a final list of candidate features.

[0043] The final list of candidate features can be analyzed according to additional feature selection techniques to determine one or more candidate feature signatures (e.g., groups or series of operating parameters that can be used to predict a metric indicative of the success of a maintenance cycle performed on an asset). Any suitable computational technique can be used to identify the candidate feature signatures using any feature selection technique, such as filtering, wrapping, and / or embedding. In an example, one or more candidate feature signatures can be selected according to a filtering method. Filtering methods include, for example, Pearson correlation, linear discriminant analysis, analysis of variance (ANOVA), chi-square test, combinations thereof, and the like. Feature selection according to filtering methods is independent of any machine learning algorithm. Instead, features can be selected based on scores in various statistical tests to correlate with an outcome variable (e.g., feature score).

[0044] One or more candidate feature signatures can be selected based on a wrapping method. A wrapping method can be configured to use a subset of features and train a machine learning model using the feature subset. Based on inferences drawn from a previous model, features can be added to and / or removed from the subset. Wrapping methods include, for example, forward feature selection, backward feature elimination, recursive feature elimination, combinations thereof, and the like. In an example, forward feature selection can be used to identify one or more candidate feature signatures. Forward feature selection is an iterative method that begins with no features in the machine learning model. In each iteration, the feature that best improves the model is added until adding the new variable does not improve the performance of the machine learning model. In an example, backward elimination can be used to identify one or more candidate feature signatures. Backward elimination is an iterative method that begins with all features in the machine learning model. In each iteration, the least significant feature is removed until no improvement is observed when the feature is removed. In an example, recursive feature elimination can be used to identify one or more candidate feature signatures. Recursive feature elimination is a greedy optimization algorithm that aims to find the best-performing feature subset. Recursive feature elimination repeatedly creates a model, retaining the best-performing or worst-performing feature at each iteration. Recursive feature elimination constructs the next model with the remaining features until all features are exhausted. Recursive feature elimination then ranks the features based on the order in which they were eliminated.

[0045] In an example, one or more candidate feature signatures can be selected based on an embedding method. The embedding method combines the qualities of the filtering method and the wrapping method. The embedding method includes, for example, the least absolute shrinkage and selection operator (LASSO) and ridge regression, which implement a penalty function to reduce overfitting. For example, LASSO regression performs L1 regularization, which adds a penalty equal to the absolute value of the coefficient size, and ridge regression performs L2 regularization, which adds a penalty equal to the square of the coefficient size.

[0046] After the automated machine learning module 230 has generated a feature set based on feature selection 231, model selection 232, and parameter selection 233, the automated machine learning module 230 may generate a machine learning-based classification model 234 based on the feature set. A machine learning-based classification model may refer to a complex mathematical model for classifying data generated using machine learning techniques. In one example, this machine learning-based classifier may include a graph of support vectors representing boundary features. For example, the boundary features may be selected from and / or represent the highest-ranked features in the feature set.

[0047] In an example, the automated machine learning module 230 can construct machine learning-based classification models 234A-234N for one or more feature scores or one or more feature score ranges using a feature set extracted from the training dataset 220. In some examples, the machine learning-based classification models 234A-234N can be combined into a single machine learning-based classification model 234. Similarly, the machine learning-based classifier 234 can represent a single classifier containing a single or multiple machine learning-based classification models 234A-234N and / or multiple classifiers containing a single machine learning-based classification model 234 or multiple machine learning-based classification models 234A-234N.

[0048] The extracted features (e.g., one or more candidate features and / or candidate feature signatures derived from the final list of candidate features) can be combined in a classification model trained using a machine learning method, such as discriminant analysis; decision trees; nearest neighbor (NN) algorithms (e.g., k-NN models, replicator NN models, etc.); statistical algorithms (e.g., Bayesian networks, etc.); clustering algorithms (e.g., k-means, mean shift, etc.); neural networks (e.g., reservoir networks, artificial neural networks, etc.); support vector machines (SVMs); logistic regression algorithms; linear regression algorithms; Markov models or chains; principal component analysis (PCA) (e.g., for linear models); multilayer perceptron (MLP) ANNs (e.g., for nonlinear models); replicating reservoir networks (e.g., for nonlinear models, typically used for time series); random forest classification; combinations thereof, etc. The resulting machine learning-based classifier 234 can include a decision rule or mapping that uses the expression levels of features in the candidate feature signature to predict a score / metric indicating the degree of success of a maintenance cycle performed on an asset based on operational data of the asset.

[0049] Candidate feature signatures and the machine learning-based classifier 234 can be used to predict a score / metric indicating the success of a maintenance cycle in a test dataset. In one example, the results of each test include a confidence level corresponding to the likelihood or probability that the corresponding test predicts a score / metric indicating the success of the maintenance cycle, or predicts a score / metric within at least a certain range. The confidence level can be a value between zero and one, representing the likelihood that the corresponding test will be associated with a score / metric indicating the success of the maintenance cycle. In one example, when there are two or more states (e.g., two or more operational assessments), the confidence level can correspond to a value p, which indicates the likelihood that a particular test will be associated with a first state. In this case, the value 1-p can indicate the likelihood that a particular test will be associated with a second state. In general, when there are more than two states, multiple confidence levels can be provided for each test and each candidate feature signature. The highest-performing candidate feature signature can be determined by comparing the results obtained for each test with a known maintenance cycle metric (e.g., an operational parameter) for each test. In general, the highest-performing candidate feature signature will have results that closely match the known maintenance cycle metric (e.g., an operational parameter).

[0050] The highest performing candidate feature signatures may be used to predict a metric indicating the degree of success of a maintenance cycle for the asset. For example, operational parameter data for a potential asset including a plurality of operational parameters may be determined / received. The operational parameter data for the potential asset may be provided to a machine learning-based classifier 234, which may predict a metric indicating the degree of success of the maintenance cycle based on the highest performing candidate feature signatures. For example, the metric may include a value (e.g., 0-10) indicating a level of success associated with the maintenance cycle. The value may be compared to a threshold (e.g., 1-10) indicating whether the maintenance cycle was successful. For example, if the metric is above the threshold, then the maintenance cycle may be determined to be unsuccessful. If the metric is below the threshold, then the maintenance cycle may be determined to be successful.

[0051] Figure 3 A flow chart is shown of an example training method 300 for generating a machine learning-based classifier 234 using the automated machine learning module 230. The automated machine learning module 230 can implement supervised, unsupervised, and / or semi-supervised (e.g., reinforcement-based) machine learning-based classification models 234A-234N. Figure 3 As shown, method 300 is an example of a supervised learning method; variations of this example of a training method are discussed below, however, other training methods can be similarly implemented to train unsupervised and / or semi-supervised machine learning models.

[0052] At step 310, the training method 300 may determine (e.g., access, receive, retrieve, etc.) operational data (e.g., one or more operational data sets) for one or more assets (e.g., transformers). The operational data may include one or more data sets, each of which may be associated with a time series of data. For example, a time series of data may be determined for multiple operating parameters of the asset, where the time series may include one or more time periods. Analysis may be performed on each time period of the data, where the data may be transformed based on the analysis of each time period of the data and provided as input data for determining / generating a predictive model. The analysis of each time period of the data may involve various sources across multiple platforms within a single utility, though some overlap in subject matter is contemplated. The analysis may include studies related to work order preventive / corrective maintenance, transformer oil dissolved gas analysis (DGA), tap controller oil DGA, inspection data, or equipment nameplate data at varying temporal resolutions. In an example, the study may include studies related to historical data, including historical sensor data derived from one or more sensors communicating with components of the asset and / or historical field test data derived from one or more field tests performed on components of the asset. As one example, each dataset may include a list of predetermined features for the tag. As another example, each dataset may include a feature score for the tag associated with one or more sets of operating parameters for the asset. The feature score may include a metric indicating the degree of success of a maintenance cycle performed on the asset based on the study or one or more sets of operating parameters.

[0053] The training method 300 may generate a training dataset and a test dataset at step 320. The training dataset and the test dataset may be generated by randomly assigning labeled feature data (e.g., labeled feature scores) associated with each feature associated with the operational data (e.g., operational parameters) to the training dataset or the test dataset. In some embodiments, the assignment of labeled feature data (e.g., labeled feature scores) associated with each feature (e.g., operational parameters) may not be completely random. In an example, only the labeled feature data (e.g., labeled feature scores) for a particular study may be used to generate the training dataset and the test dataset. As an example, a majority of the labeled feature data (e.g., labeled feature scores) for a particular study may be used to generate the training dataset. For example, 75% of the labeled feature data (e.g., labeled feature scores) for a particular study may be used to generate the training dataset, and 25% may be used to generate the test dataset. As another example, only the labeled feature data (e.g., labeled feature scores) for a particular study may be used to generate the training dataset and the test dataset.

[0054] The training method 300 may determine (e.g., extract, select, etc.) one or more features at step 330 that may be used by, for example, a classifier to distinguish between different sets of operating parameters to determine one or more feature scores or one or more feature score ranges associated with the one or more features. The one or more features may include a data set in an operating data set associated with a set of operating parameters. For example, the training method 300 may determine a feature set from the operating data. As another example, a feature set may be determined from operating data from a study that is different from the study associated with the labeled feature data (e.g., labeled feature scores) of the training data set and the test data set. In other words, operating data from different studies (e.g., a selected operating data set) may be used for feature determination rather than for training a machine learning model. In an example, the training data set may be used in conjunction with operating data from different studies to determine one or more features. The operating data from different studies may be used to determine an initial feature set, which may be further simplified using the training data set.

[0055] The training method 300 can train one or more machine learning models using the one or more features at step 340. As an example, the machine learning models can be trained using supervised learning. As another example, other machine learning techniques can be employed, including unsupervised learning and semi-supervised learning. The machine learning model trained at step 340 can be selected based on different criteria depending on the problem to be solved and / or the data available in the training dataset. For example, machine learning classifiers can be subject to varying degrees of bias. Thus, more than one machine learning model can be trained at step 340 and optimized, refined, and cross-validated at step 350.

[0056] The training method 300 may select one or more machine learning models to construct a prediction model at step 360 (e.g., a machine learning classifier). The prediction model may be evaluated using a test data set. At step 370, the prediction model may analyze the test data set and generate a classification value (e.g., a feature score) and / or a predicted value (e.g., a feature score). At step 380, the classification value and / or predicted value (e.g., a feature score) may be evaluated to determine whether these values ​​have reached the desired level of accuracy. The performance of the prediction model may be evaluated in a variety of ways based on the number of true positive, false positive, true negative, and / or false negative classifications of a plurality of data points indicated by the prediction model. For example, a false positive of a prediction model may refer to the number of times the prediction model misclassifies / scores a set of operating parameters. Conversely, a false negative of a prediction model may refer to the number of times the machine learning model determines that a classification value (e.g., a feature score) is not associated with a set of operating parameters when, in fact, the classification value (e.g., a feature score) is associated with the set of operating parameters. True negatives and true positives may refer to the number of times the prediction model correctly classifies / scores one or more sets of operating parameters. Related to these measurements are the concepts of recall and precision. Typically, recall refers to the ratio of true positives to the sum of true positives and false negatives, which quantifies the sensitivity of a predictive model. Similarly, precision refers to the ratio of true positives to the sum of true positives and false positives.

[0057] When this desired level of accuracy is achieved, the training phase ends and the predictive model can be output at step 390; however, when the desired level of accuracy is not achieved, subsequent iterations of the training method 300 can be performed at step 310 in a modified form, for example, considering a larger set of operational data.

[0058] Figure 4 A block diagram of an environment 400 is shown, comprising a non-limiting example of a computing device 401 and a server 402 connected via a network 404. In one aspect, some or all of the steps of any of the described methods can be performed on a computing device as described herein. The computing device 401 can include one or more computers configured to store one or more of the training modules 230, the training data 220 (e.g., operational data 424), etc. The server 402 can include one or more computers configured to store operational data 424 (e.g., selected parameter set data). Multiple servers 402 can communicate with the computing device 401 via the network 404.

[0059] Computing device 401 and server 402 can be digital computers that, in terms of hardware architecture, typically include a processor 408, a memory system 410, an input / output (I / O) interface 412, and a network interface 414. These components (408, 410, 412, and 414) are communicatively coupled via a local interface 416. Local interface 416 can be, for example, but not limited to, one or more buses or other wired or wireless connections, as known in the art. Local interface 416 can have additional components, which are omitted for simplicity, such as controllers, buffers (caches), drivers, repeaters, and receivers, to enable communication. In addition, the local interface can include address, control, and / or data connections to enable appropriate communication between the aforementioned components.

[0060] Processor 408 may be a hardware device for executing software, particularly software stored in memory system 410. Processor 408 may be any custom or commercially available processor, a central processing unit (CPU), a secondary processor among several processors associated with computing device 401 and server 402, a semiconductor-based microprocessor (in the form of a microchip or chipset), or generally any device for executing software instructions. When computing device 401 and / or server 402 is operating, processor 408 may be configured to execute software stored in memory system 410, to transfer data to and from memory system 410, and generally control the operation of computing device 401 and server 402 according to the software.

[0061] The I / O interface 412 may be used to receive user input from one or more devices or components and / or provide system output to one or more devices or components. User input may be provided, for example, by a keyboard and / or a mouse. System output may be provided via a display device and a printer (not shown). The I / O interface 412 may include, for example, a serial port, a parallel port, a small computer system interface (SCSI), an infrared (IR) interface, a radio frequency (RF) interface, and / or a universal serial bus (USB) interface.

[0062] The network interface 414 may be used to transmit and receive from the computing device 401 and / or the server 402 on the network 404. The network interface 414 may include, for example, a 10BaseT Ethernet adapter, a 100BaseT Ethernet adapter, a LAN PHY Ethernet adapter, a token ring adapter, a wireless network adapter (e.g., WiFi, cellular, satellite), or any other suitable network interface device. The network interface 414 may include address, control, and / or data connections to enable appropriate communications on the network 404.

[0063] The memory system 410 may include any one or a combination of volatile memory elements (e.g., random access memory (RAM, such as DRAM, SRAM, SDRAM, etc.)) and non-volatile memory elements (e.g., ROM, hard drive, tape, CDROM, DVDROM, etc.). In addition, the memory system 410 may include electronic, magnetic, optical, and / or other types of storage media. It should be noted that the memory system 410 may have a distributed architecture in which various components are remote from each other but accessible by the processor 408.

[0064] The software in the memory system 410 may include one or more software programs, each of which includes an ordered listing of executable instructions for implementing logical functions. Figure 4 In the example of FIG, the software in the memory system 410 of the computing device 401 may include the training module 230 (or subcomponents thereof), the training data 220, and a suitable operating system (O / S) 418. Figure 4 In the example of FIG, the software in the memory system 410 of the server 402 may include the training data 210 and a suitable operating system (O / S) 418. The operating system 418 generally controls the execution of other computer programs and provides scheduling, input-output control, file and data management, memory management, communication control, and related services.

[0065] For purposes of illustration, application programs and other executable program components (e.g., operating system 418) are shown herein as discrete blocks, but it should be appreciated that these programs and components may reside at different times in different storage components of computing device 401 and / or server 402. Implementations of training module 230 may be stored on or transmitted across some form of computer-readable media. Any of the disclosed methods may be performed by computer-readable instructions embodied on computer-readable media. Computer-readable media can be any available media that can be accessed by a computer. By way of example, and not intended to be limiting, computer-readable media may include "computer storage media" and "communication media." "Computer storage media" may include volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information, such as computer-readable instructions, data structures, program modules, or other data. Exemplary computer storage media may include RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disks (DVDs) or other optical storage devices, cassettes, magnetic tape, magnetic disk storage devices or other magnetic storage devices, or any other media that can be used to store the desired information and that can be accessed by a computer.

[0066] In an example, the training module 230 can be configured to perform the method 500 as follows: Figure 5As shown. Method 500 can be performed in whole or in part by a single computing device, multiple electronic devices, etc. Method 500 may include, at 510, determining operational data associated with a plurality of operating parameters. The operational data includes one or more data sets based on a plurality of operating parameters of an asset (e.g., a transformer). Each of the one or more data sets may include data indicating a time series of data associated with the plurality of operating parameters of the asset, wherein the time series of data may include one or more time periods of the data. The plurality of operating parameters may include one or more of the following: a power parameter data set, a voltage parameter data set, a current parameter data set, a capacity parameter data set, a heat parameter data set, a cooling pipe parameter data set, a fuel tank parameter data set, a sunlight duration parameter data set, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density, or an air temperature parameter data set. The plurality of operating parameters may include one or more groups of operating parameters. Each group of operating parameters in the plurality of operating parameters may be labeled according to a predefined feature score from a plurality of predefined features. The feature score may include a metric indicating a degree of success of a maintenance cycle performed on the asset based on the one or more groups of operating parameters.

[0067] In one example, determining operational data associated with multiple operational parameters may further include determining a time series of data associated with the multiple operational parameters of an asset, wherein the time series includes one or more time periods. Analysis may be performed on each time period of the data in the one or more time periods. Operational data may be generated based on the analysis of each time period of the data, wherein the operational data includes a data set associated with each time period. As another example, determining operational data associated with multiple operational parameters may further include determining one or more operational data sets including at least one of the multiple operational parameters based on the multiple operational parameters. Operational data may be generated based on the one or more operational data sets. As another example, determining operational data associated with multiple operational parameters may further include determining a baseline characteristic score for each group of operational parameters in the multiple operational parameters. The baseline characteristic score for each group of operational parameters may then be labeled as a characteristic score associated with each group of operational parameters. Operational data may be generated based on the labeled baseline characteristics for each group of operational parameters.

[0068] Method 500 may include determining a plurality of feature scores of a prediction model based on the operational data at step 520. In an example, determining the plurality of feature scores of the prediction model based on the operational data may include: determining, from the operational data, feature scores associated with two or more operational data sets in a plurality of operational data sets as a first candidate feature score set; determining, from the operational data, feature scores associated with the first candidate feature score set that satisfy a first threshold score as a second candidate feature score set; and determining, from the operational data, feature scores of the second candidate feature score set that satisfy a second threshold score as a third candidate feature score set, wherein the plurality of feature scores includes the third candidate feature score set.

[0069] Method 500 may include training a predictive model based on the first portion of the operating parameter data according to the plurality of feature scores at step 530. Training the predictive model based on the first portion of the operating parameter data according to the plurality of feature scores results in determining a feature signature indicative of a feature score associated with each set of operating parameters.

[0070] The method 500 may include testing the predictive model based on the second portion of the operational data at step 540. The method 500 may include outputting the predictive model based on the testing at 550. The predictive model may be configured to output a prediction indicating a degree of success or level of success of a maintenance cycle performed on the asset. For example, the predictive model may be configured to output a predictive score associated with a maintenance cycle performed on the asset. For example, the predictive score may include a value (e.g., 1-10) indicating a degree of success / level of success of the maintenance cycle. The predictive score may be compared to a threshold / score (e.g., 1-10) to determine whether the maintenance cycle was successful. If the predictive score is above the threshold, then the maintenance cycle may be determined to be unsuccessful. If the predictive score is below the threshold, then the maintenance cycle may be determined to be successful.

[0071] In an example, the training module 230 can be configured to perform the method 600 as follows: Figure 6As shown. Method 600 can be performed in whole or in part by a single computing device, multiple electronic devices, or the like. Method 600 may include, at step 610, receiving operating parameter data associated with a plurality of operating parameters of an asset (e.g., a transformer). The plurality of operating parameters may be determined during operation of the asset. The plurality of operating parameters may include one or more of the following: power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, fuel tank parameters, sunlight duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density, or air temperature parameters. Method 600 may include, at step 620, providing the operating parameter data to a predictive model. Method 600 may include, at step 630, determining a prediction score associated with a maintenance cycle performed on the asset based on the predictive model. The prediction score may include a metric indicating a degree of success of the maintenance cycle performed on the asset. For example, the metric may include a value (e.g., 1-10) indicating a degree of success / level of success of the maintenance cycle. The prediction score may be compared to a threshold / score (e.g., 1-10) to determine whether the maintenance cycle was successful. If the prediction score is above the threshold, the maintenance cycle may be determined to be unsuccessful. If the prediction score is below a threshold, the maintenance cycle can be determined to be successful.

[0072] Training a predictive model may include: determining operational data associated with multiple operational parameters of an asset, wherein the multiple operational parameters include one or more groups of operational parameters, wherein each group of operational parameters in the multiple operational parameters is labeled according to a feature score; determining multiple feature scores for the predictive model based on the operational data; training the predictive model based on a first portion of the operational data based on the multiple feature scores; testing the predictive model based on a second portion of the operational data; and outputting the predictive model based on the testing.

[0073] The operational data may include one or more data sets, wherein each of the one or more data sets includes data indicative of a time series of data associated with the asset.

[0074] Determining operational data associated with a plurality of operational parameters may further include determining a time series of data associated with the asset, wherein the time series includes one or more time periods. Analysis may be performed on each time period of the data in the one or more time periods of the data. Operational data may be generated based on the analysis of each time period of the data. As another example, determining operational data associated with a plurality of operational parameters may further include determining, based on the plurality of operational parameters, one or more operational data sets including at least one operational parameter of the plurality of operational parameters. As another example, determining operational data associated with a plurality of operational parameters may further include determining a baseline characteristic score for each set of operational parameters associated with the plurality of operational parameters. The baseline characteristic score for each set of operational parameters may then be labeled as a characteristic score associated with each set of operational parameters.

[0075] Determining multiple feature scores of a prediction model based on operational data may include: determining, from the operational data, feature scores associated with two or more operational parameter data sets from a plurality of different operational parameter data sets as a first candidate feature score set; determining, from the operational data, feature scores of the first candidate feature score set that satisfy a first threshold score as a second candidate feature score set; and determining, from the operational data, feature scores of the second candidate feature score set that satisfy a second threshold score as a third candidate feature score set, wherein the multiple feature scores include the third candidate feature score set.

[0076] Training the predictive model based on the first portion of the operational data results in determining a feature signature indicative of a feature score associated with each set of operating parameters according to the plurality of feature scores.

[0077] Embodiment 1: A method comprising: determining operational data associated with a plurality of operational parameters associated with an asset, wherein the plurality of operational parameters include one or more groups of operational parameters, and wherein each group of operational parameters in the one or more groups of operational parameters is labeled according to a feature score; determining a plurality of feature scores of a prediction model based on the operational data; training the prediction model based on a first portion of the operational data according to the plurality of feature scores; testing the prediction model based on a second portion of the operational data; and outputting the prediction model based on the testing.

[0078] Embodiment 2: The embodiment according to any of the preceding embodiments, wherein the operational data comprises one or more data sets, wherein each of the one or more data sets comprises data indicative of a time series of data associated with the asset.

[0079] Embodiment 3: An embodiment according to any one of the preceding embodiments, wherein the multiple operating parameters include one or more of the following: power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, oil tank parameters, sunshine duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density or air temperature parameters.

[0080] Embodiment 4: An embodiment according to any of the preceding embodiments, wherein the asset comprises a transfer.

[0081] Embodiment 5: The embodiment according to any one of the preceding embodiments, wherein determining the operational data associated with the plurality of operational parameters comprises retrieving the operational data from a public data source.

[0082] Embodiment 6: An embodiment according to any one of the preceding embodiments, wherein determining the operational data associated with the multiple operational parameters includes: determining a time series of data associated with the multiple operational parameters associated with the asset, wherein the time series includes one or more time periods; performing analysis on each time period of the data in the one or more time periods of the data; and generating the operational data based on the analysis of each time period of the data, wherein the operational data includes a data set associated with each time period.

[0083] Example 7: An embodiment according to any one of the preceding embodiments, wherein determining the operation data associated with the multiple operation parameters includes: determining one or more operation data sets associated with at least one operation parameter among the multiple operation parameters based on the multiple operation parameters; and generating the operation data based on the one or more operation data sets.

[0084] Example 8: An embodiment according to any one of the preceding embodiments, wherein determining the operational data associated with the multiple operational parameters includes: determining a baseline characteristic score for each group of operational parameters in the multiple operational parameters; marking the baseline characteristic score for each group of operational parameters in the multiple operational parameters as the characteristic score associated with each group of operational parameters; and generating the operational data based on the marked baseline characteristic score.

[0085] Example 9: An embodiment according to any one of the preceding embodiments, wherein determining the multiple feature scores of the prediction model based on the operational data includes: determining, from the operational data, feature scores associated with two or more of the multiple operational data sets as a first candidate feature score set; determining, from the operational data, feature scores associated with the first candidate feature score set that meet a first threshold score as a second candidate feature score set; and determining, from the operational data, feature scores associated with the second candidate feature score set that meet a second threshold score as a third candidate feature score set, wherein the multiple feature scores include the third candidate feature score set.

[0086] Embodiment 10: The embodiment according to any one of the preceding embodiments, wherein the feature score comprises a metric indicative of a degree of success of a maintenance cycle performed on the asset based on the one or more sets of operating parameters.

[0087] Embodiment 11: An embodiment according to any of the preceding embodiments, wherein training the predictive model based on the first portion of the operational data results in determining a feature signature indicative of the feature scores associated with each set of operating parameters based on the plurality of feature scores.

[0088] Embodiment 12: The embodiment according to any one of the preceding embodiments, wherein the predictive model is configured to output a prediction indicating a degree of success of a maintenance cycle performed on the asset.

[0089] Embodiment 13: The embodiment according to any one of the preceding embodiments, wherein the predictive model is configured to output a predictive score associated with a maintenance period performed on the asset.

[0090] Embodiment 14: The embodiment of embodiment 13, further comprising determining a prediction indicating success of the maintenance cycle based on the prediction score satisfying a threshold.

[0091] Embodiment 15: The embodiment of embodiment 13, further comprising determining a prediction indicating that the maintenance cycle was unsuccessful based on the prediction score satisfying a threshold.

[0092] Embodiment 16: A method comprising: receiving operating parameter data comprising a plurality of operating parameters of an asset, wherein the plurality of operating parameters are determined during an analysis of one or more operations performed by the asset; providing the operating parameter data to a predictive model; and determining a predictive score associated with a maintenance cycle performed on the asset based on the predictive model.

[0093] Example 17: An embodiment according to Example 16, wherein the multiple operating parameters include one or more of the following: power parameters, voltage parameters, current parameters, capacity parameters, heat parameters, cooling pipe parameters, oil tank parameters, sunshine duration parameters, transformer height, manufacturing date, manufacturer data, installation date, vehicle traffic density or air temperature parameters.

[0094] Embodiment 18: The embodiment according to any one of Embodiments 16 to 17, wherein the asset comprises a transformer.

[0095] Embodiment 19: According to any one of embodiments 16 to 18, it also includes training the prediction model.

[0096] Example 20: An embodiment according to any one of Examples 16 to 19, wherein training the prediction model includes: determining operational data associated with the multiple operational parameters of the asset, wherein the multiple operational parameters include one or more groups of operational parameters, wherein each group of operational parameters in the one or more groups of operational parameters is labeled according to a feature score; determining multiple feature scores of the prediction model based on the operational data; training the prediction model based on a first portion of the operational data according to the multiple feature scores; testing the prediction model based on a second portion of the operational data; and outputting the prediction model based on the test.

[0097] Embodiment 21: The embodiment of embodiment 20, wherein the operational data comprises one or more data sets, wherein each of the one or more data sets comprises data indicative of a time series of data associated with the asset.

[0098] Embodiment 22: The embodiment of embodiments 20-21, wherein determining the operational data associated with the plurality of operational parameters comprises retrieving the operational data from a common data source.

[0099] Example 23: An embodiment according to Examples 20 to 22, wherein determining the operational data associated with the multiple operational parameters includes: determining a time series of data associated with the multiple operational parameters associated with the asset, wherein the time series includes one or more time periods; performing analysis on each time period of the data in the one or more time periods of the data; and generating the operational data based on the analysis of each time period of the data, wherein the operational data includes a data set associated with each time period.

[0100] Example 24: An embodiment according to Examples 20 to 23, wherein determining the operation data associated with the multiple operation parameters includes: determining one or more operation data sets associated with at least one operation parameter of the multiple operation parameters based on the multiple operation parameters; and generating the operation data based on the one or more operation data sets.

[0101] Example 25: An embodiment according to Examples 20 to 24, wherein determining the operational data associated with the multiple operational parameters includes: determining a baseline characteristic score for each group of operational parameters in the multiple operational parameters; marking the baseline characteristic score for each group of operational parameters in the multiple operational parameters as the characteristic score associated with each group of operational parameters; and generating the operational data based on the marked baseline characteristic score.

[0102] Example 26: An embodiment according to Examples 20 to 25, wherein determining the multiple feature scores of the prediction model based on the operation data includes: determining feature scores associated with two or more operation parameter data sets in a plurality of parameter data sets from the operation data as a first candidate feature score set; determining feature scores in the first candidate feature score set that meet a first threshold score from the operation data as a second candidate feature score set; and determining feature scores in the second candidate feature score set that meet a second threshold score from the operation data as a third candidate feature score set, wherein the multiple feature scores include the third candidate score set.

[0103] Embodiment 27: The embodiment of embodiments 20 to 26, wherein the feature score comprises a metric indicative of a degree of success of a maintenance cycle performed on the asset based on the one or more sets of operating parameters.

[0104] Embodiment 28: An embodiment according to embodiments 20 to 27, wherein training the predictive model based on the first portion of the operational data results in determining a feature signature indicative of the feature scores associated with each set of operating parameters based on the plurality of feature scores.

[0105] Embodiment 29: The embodiment of embodiments 16 to 28, further comprising determining a prediction indicating success of the maintenance cycle based on the prediction score satisfying a threshold.

[0106] Embodiment 30: The embodiment of embodiments 16 to 28, further comprising determining a prediction indicating that the maintenance cycle was unsuccessful based on the prediction score satisfying a threshold.

[0107] Although the methods and systems have been described in conjunction with preferred embodiments and specific examples, it is not intended to limit the scope to the specific embodiments set forth, as the embodiments herein are intended in all respects to be illustrative rather than restrictive.

[0108] Unless otherwise expressly stated, it is not intended that any method described herein be construed as requiring that its steps be performed in a specific order. Therefore, where a method claim does not actually recite the order in which its steps are to be followed, or where the claims or description do not otherwise expressly state that the steps are to be limited to a specific order, no inference of such order is intended. This applies to any possible non-express basis for interpretation, including: logical issues regarding the arrangement of steps or operational flow; plain meaning derived from grammatical organization or punctuation; or the number or type of embodiments described in the specification.

[0109] It will be apparent to those skilled in the art that various modifications and variations can be made without departing from the scope or spirit of the invention. Other embodiments will be apparent to those skilled in the art from consideration of the specification and practice disclosed herein. It is intended that the description and examples be considered exemplary only, with the true scope and spirit being indicated by the appended claims.

Claims

1. A method comprising: determining, by a computing device, operational data associated with a plurality of operating parameters associated with the asset, wherein the plurality of operating parameters comprises one or more groups of operating parameters, and wherein each group of operating parameters in the one or more groups of operating parameters is labeled according to a characteristic score; determining a plurality of feature scores for a predictive model based on the operational data; training the predictive model based on a first portion of the operational data according to the plurality of feature scores; testing the predictive model based on a second portion of the operational data; as well as The predictive model is output based on the testing.

2. The method of claim 1, wherein the operational data comprises one or more data sets, wherein each of the one or more data sets comprises data indicative of a time series of data associated with the plurality of operational parameters associated with the asset.

3. The method of claim 1 , wherein the plurality of operating parameters comprises one or more of: a power parameter, a voltage parameter, a current parameter, a capacity parameter, a heat parameter, a cooling pipe parameter, a tank parameter, a sunshine duration parameter, a transformer height, a manufacturing date, a manufacturer's data, an installation date, a vehicle traffic density, or an air temperature parameter. The method of claim 1 , wherein the asset comprises a transformer.

5. The method of claim 1 , wherein determining the operational data associated with the plurality of operational parameters comprises: determining a time series of data associated with the plurality of operating parameters associated with the asset, wherein the time series includes one or more time periods; performing analysis on each time period of the data in the one or more time periods of the data; and The operational data is generated based on the analysis of each time period of the data, wherein the operational data includes a data set associated with each time period.

6. The method of claim 1 , wherein determining the operational data associated with the plurality of operational parameters comprises: determining a baseline characteristic score for each set of operating parameters in the plurality of operating parameters; labeling the baseline characteristic score for each set of operating parameters in the plurality of operating parameters as the characteristic score associated with each set of operating parameters; as well as The operational data is generated based on the baseline characteristic scores of the markers.

7. The method of claim 1 , wherein determining the plurality of feature scores of the predictive model based on the operational data comprises: determining, from the operational data, feature scores associated with two or more operational data sets among the plurality of operational data sets as a first candidate feature score set; Determining, from the operational data, feature scores associated with the first candidate feature score set that meet a first threshold score as a second candidate feature score set; as well as determining, from the operation data, a feature score associated with the second candidate feature score set that satisfies a second threshold score as a third candidate feature score set; The plurality of feature scores include the third candidate feature score set.

8. The method of claim 1, wherein the predictive model is configured to output a predictive score indicating a degree of success of a maintenance cycle performed on the asset. 9 . The method of claim 8 , further comprising determining a prediction indicating success of the maintenance cycle based on the prediction score satisfying a threshold.

10. The method of claim 8, further comprising determining a prediction indicating that the maintenance cycle was unsuccessful based on the prediction score satisfying a threshold.

11. A method comprising: receiving, at a computing device, operating parameter data comprising a plurality of operating parameters of an asset, wherein the plurality of operating parameters are determined during analysis of one or more operations performed by the asset; providing the operating parameter data to a predictive model; as well as A prediction score associated with a maintenance cycle performed on the asset is determined based on the prediction model.

12. The method of claim 11, wherein the plurality of operating parameters comprises one or more of: a power parameter, a voltage parameter, a current parameter, a capacity parameter, a heat parameter, a cooling pipe parameter, a tank parameter, a sunshine duration parameter, a transformer height, a manufacturing date, a manufacturer's data, an installation date, a vehicle traffic density, or an air temperature parameter.

13. The method of claim 11, wherein the asset comprises a transformer. The method of claim 11 , further comprising training the prediction model.

15. The method of claim 14, wherein training the prediction model comprises: determining operational data associated with the plurality of operating parameters of the asset, wherein the plurality of operating parameters comprises one or more groups of operating parameters, and wherein each group of operating parameters in the one or more groups of operating parameters is labeled according to a feature score; determining a plurality of feature scores for the predictive model based on the operational data; training the predictive model based on a first portion of the operational data according to the plurality of feature scores; testing the predictive model based on a second portion of the operational data; as well as The predictive model is output based on the testing.

16. The method of claim 15, wherein determining the operational data associated with the plurality of operational parameters comprises: determining a time series of data associated with the plurality of operating parameters associated with the asset, wherein the time series includes one or more time periods; performing analysis on each time period of the data in the one or more time periods of the data; and The operational data is generated based on the analysis of each time period of the data, wherein the operational data includes a data set associated with each time period.

17. The method of claim 15, wherein determining the operational data associated with the plurality of operational parameters comprises: determining a baseline characteristic score for each set of operating parameters in the plurality of operating parameters; labeling the baseline characteristic score for each set of operating parameters in the plurality of operating parameters as the characteristic score associated with each set of operating parameters; as well as The operational data is generated based on the baseline characteristic scores of the markers.

18. The method of claim 15, wherein determining the plurality of feature scores of the predictive model based on the operational data comprises: determining, from the operational data, feature scores associated with two or more operational data sets among the plurality of operational data sets as a first candidate feature score set; Determining, from the operational data, feature scores associated with the first candidate feature score set that meet a first threshold score as a second candidate feature score set; as well as determining, from the operation data, a feature score associated with the second candidate feature score set that satisfies a second threshold as a third candidate feature score set; The plurality of feature scores include the third candidate feature score set.

19. The method of claim 11, further comprising determining a prediction indicating success of the maintenance cycle based on the prediction score satisfying a threshold.

20. The method of claim 11, further comprising determining a prediction indicating that the maintenance cycle was unsuccessful based on the prediction score satisfying a threshold.

Images