Classifications

• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
  • G06Q50/06—Energy or water supply
• • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
  • F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
  • F04B—POSITIVE-DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS
  • F04B49/00—Control, e.g. of pump delivery, or pump pressure of, or safety measures for, machines, pumps, or pumping installations, not otherwise provided for, or of interest apart from, groups F04B1/00 - F04B47/00
  • F04B49/06—Control using electricity
  • F04B49/065—Control using electricity and making use of computers
• • F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
  • F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
  • F04B—POSITIVE-DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS
  • F04B51/00—Testing machines, pumps, or pumping installations
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/044—Recurrent networks, e.g. Hopfield networks
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/04—Architecture, e.g. interconnection topology
  • G06N3/044—Recurrent networks, e.g. Hopfield networks
  • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
  • G06N3/09—Supervised learning
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
  • G06N3/00—Computing arrangements based on biological models
  • G06N3/02—Neural networks
  • G06N3/08—Learning methods
  • G06N3/098—Distributed learning, e.g. federated learning
• • G—PHYSICS
  • G06—COMPUTING OR CALCULATING; COUNTING
  • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
  • G06Q10/00—Administration; Management
  • G06Q10/20—Administration of product repair or maintenance
• • G—PHYSICS
  • G05—CONTROLLING; REGULATING
  • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
  • G05B23/00—Testing or monitoring of control systems or parts thereof
  • G05B23/02—Electric testing or monitoring
  • G05B23/0205—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults
  • G05B23/0259—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterized by the response to fault detection
  • G05B23/0283—Predictive maintenance, e.g. involving the monitoring of a system and, based on the monitoring results, taking decisions on the maintenance schedule of the monitored system; Estimating remaining useful life [RUL]

Definitions

• the present disclosure generally relates to artificial intelligence and machine learning systems, and more particularly, to methods and systems for distributed learning for multi-label classification models for industrial equipment repair and maintenance, such as oil pump preventive management.
• a computing device a non-transitory computer readable storage medium, and a method are provided for developing and improving a predictive model for asset failure where only the models are shared among different sites.
• the present invention provides a computer implemented method of predicting failure of assets includes partitioning features of the assets into static features, semi-static features and dynamic features and forming cohorts of the assets based on the static features and the semi-static features.
• the method further comprises generating a local model at a local site for each of the cohorts and training the local model on local data for each of the cohorts for each failure type.
• the local model is shared with a central database.
• a global model is created based on an aggregation of a plurality of the local models from a plurality of the local sites. At each of the plurality of local sites, one of the global model and the local model is chosen for each of the cohorts. The chosen model operates on local data to predict the failure of one or more of the assets belonging to this cohort.
• the present invention provides a method further comprising generating a template model for creating each of the local models for each of the cohorts.
• the present invention provides a method further comprising pooling the local models from each of the plurality of local sites into a pool of local models and determining the performance of the global model and a selected one of the local models from the pool of local models.
• the present invention provides a method further comprising updating the global model based on an average of each of the local models in the pool of local models for each of the cohorts.
• each of the local models in the pool of local models is weighted based on an average number of assets of the local site that provided the local model to the pool of local models.
• a computer implemented method for predicting failure of assets comprises partitioning features of the assets into static features, semi-static features, and dynamic features and forming cohorts of the assets based on the static features and the semi-static features.
• a local model is generated at a local site for each of the cohorts, and the local model is shared with a central database.
• a pool of local models is created from a plurality of the local models from a respective plurality of the local sites.
• a global model is created based on an aggregation of the plurality of the local models from the plurality of the local sites.
• one of the global models and one of the pluralities of local models from the pool of local models is chosen for each of the cohorts. The chosen model operates on local data to predict failure of the assets.
• a computer implemented method for prediction failure of assets comprises partitioning features of the assets into static features, semi-static features and dynamic features and forming cohorts of the assets based on the static features and the semi-static features.
• a local model is generated at a local site for each of the cohorts and a global model is created for each of the cohorts.
• At each of a plurality of local sites, one of the global model and the local model is chosen for each of the cohorts. The chosen model operates on local data to predict the failure of the assets belong to the cohost where the model is developed.
• FIG. 1 is a representation of model deployment based on feature grouping, according to an embodiment of the present disclosure.
• FIG. 2 is a representation of a system architecture of an asset failure prediction engine after deployment of federated models, illustrating model tuning and model updating, consistent with an illustrative embodiment.
• FIG. 3 illustrates a method for local model selection for a cohost at a local site, consistent with an illustrative embodiment.
• FIG. 4 illustrates a method for local model selection, without sharing of local models from other sites, consistent with an illustrative embodiment.
• FIG. 5 illustrates a model template for future cohost multiple-label prediction, consistent with an illustrative embodiment.
• FIG. 6 illustrates an architecture model for mismatch and tuning at a local site, consistent with an illustrative embodiment.
• FIG. 7 is a flow chart illustrating acts involved with setting up an asset failure prediction engine, consistent with an illustrative embodiment.
• FIG. 8 is a flow chart illustrating acts involved with operating an asset failure prediction engine, consistent with an illustrative embodiment.
• FIG. 9 is a functional block diagram illustration of a computer hardware platform that can be used to implement the asset failure prediction engine of FIG. 2.
• distributed learning refers to a learning model where local data is not shared, only models (local and global) are shared among distributed sites.
• multi-learning classification refers to a classification problem where multiple labels may be assigned to each case, where they are not mutually exclusive.
• the term “cohort” refers to, instead of grouping all the assets as a single category, breaking down the assets into different groups for better analysis and prediction.
• global model refers to a model for a specific cohort, where a local site is allowed to use the global model, either due to lack of a local model, or because the local model is not as good as the global model.
• the term “local model” refers to the model used at a local site.
• model aggregator refers to an updated global model based on the collective of local models and their usage.
• model selection at a local site refers to an algorithm used to select a local model to use at the local site, where the pool of the candidate models comes from either the global model for a cohort and/or local models from other sites.
• the present disclosure generally relates to methods and systems for industrial asset management, such as oil pump management, by distributed learning.
• the methods and systems use a central approach to store a cohort model and prediction model and dispatch the centralized models to local sites as a starting point for the local site.
• the methods and systems can use a distributed approach to refine and improve the prediction model performance at a local site, resulting in local refinement by retraining the model with the same model structure and initial features.
• a centralized management of the cohort model can help ensure learning from each site will be aligned to a common or similar scenario group.
• the systems and methods of the present disclosure can provide a significant improvement in asset failure prediction management models by sharing models across multiple sites in efforts to use information learned from one site at other sites, without having to share the specific details of any specific asset failure.
• the present disclosure provides systems and methods that can perform model mismatch analysis to decide when to update the local model and generate model refinements.
• the systems and methods can apply a predictive model, including but not limited to traditional and deep learning model architecture for a prediction or classification model with sufficient model complexity to allow model integration from the different sites and to avoid loss of accuracy for other sites after global model integration.
• ESP submersible pump or electric submersible pump
• PCP progressing cavity pump
• RRL reciprocating rod lift
• Each of these pumps can represent a specific cohort.
• the types of machinery belong to same cohort sharing similar structures and functionalities.
• each oil pump model family 100 can be used to provide a set of features 102 for that model family 100.
• the set of features 102 can include static features 104, semi-static features 106, and dynamic features 108.
• a global database 110 can be maintained that includes data regarding each asset (e.g., each pump family) and the features of each asset.
• Static features 104 can include information that typically does not change for an asset, such as the asset purchase year, model number, brand, geographic location, and the like.
• Semi-static features 106 can include information that changes slowly, such as days from purchase, the days from last maintenance, number of scheduled maintenance from purchase, the average days of repair intervals, and the like.
• Dynamic features 108 include information from monitoring sensors. Typically, dynamic features 108 will provide a failure signal in a short time horizon. Certain data aggregation and transformation might be used to convert this data into usable features, such as hourly/daily average, exponential smoothness, outlier identification, missing values for a previous week, or the like.
• the assets can be divided into one or more cohorts 112.
• the cohorts are based on pump type (pump type A, pump type B, pump type C, and pump type D), but, depending on the asset of interest, the cohorts may be established based on any given asset or division of assets.
• a windmill farm may divide cohorts into different power generation components, such as bearings, inverters, storage devices, or the like.
• the cohorts 112 can be created by analyzing the static and semi-static features of the assets, where a given cohort can have similar static and/or semi-static features for a given asset.
• cohorts 112 can include the static features 104, semi-static features 106, and dynamic features 108 for that particular asset.
• a failure prediction model 114 (also referred to as a global model 114) can be built for each cohort.
• the global model 114 may be based on a previously established local model for a given asset or cohort or may be based on a model based on a similar asset.
• the global model 114 can have a deep learning model structure with a fixed architecture based on the static, semi-static, and dynamic features of the asset.
• the global models 114, based on the cohorts 112 can be deployed to each of a plurality of local sites 116.
• each deployment site 200 can select a local model 202 from a model repository 204.
• the model repository 204 can include a global model for each cohort.
• the model repository can include both a global model and at least one local model for each cohort.
• a model consolidator/aggregator 206 may be provided to match models with cohorts and to limit the overall number of models by removing similar or identical local models.
• Local mismatch analysis 208 may be performed at each deployment side 200, where the selected local model 202 can be monitored for model performance. Discrepancies between model predictions and actual asset performance may result in the reporting of a mismatch report 210 for federated analysis 212.
• This analysis is referred to as “federated analysis” because mismatch reports 210 from various local sites 200 may be analyzed together.
• the system 250 may provide a cohort update 214 or a model update 216, which can be sent to the model repository 204.
• each local site 200 can provide local model tuning 216. Details of model tuning 216 may be provided for federated analysis 212 and model updates 216 for an updated local model for the model repository 204.
• the local site 300 can receive one or more local models 302 and a global model 304 for the given cohort. If there are no models available (as either global or local models) for a given cohort (such as when a new cohort is established at a local site), then a new local model is generated at the local site. This new local model may be, for example, based on a model for a similar asset at this site or another local site.
• Two models can be selected for comparison and refinement.
• One of the selected models can be the global model for the given cohort, and the other model can be a local model from the local model pool.
• Various criteria may be used for selecting a local model from the local model pool, such as the similarity of the site providing the local model to the local site 300, the proximity of the site providing the local model to the local site 300, the performance metric of the local model, or the like.
• the local model may be selected from a local model at the local site 300 itself. Data can be applied to the two selected models, and the performance can be analyzed. The best performing model can be selected as the new local model 306 for the given cohort at the local site 300. If there is a performance mismatch between the two models, tuning can be performed to generate the new local model and this new local model can be shared to the global model repository 204 (see FIG. 2).
• the local models may not be shared due to site management or privacy requirements.
• a global model 402 may be provided to a local site 400 for a given cohort. Again, two models may be selected, where one model is the global model 204, and the other is the current version of the local model generated at the local site 400. If there is no local model currently at the local site 400, the local site 400 may create a local model in a manner similar to that discussed above with respect to FIG. 3. The models can be compared and refined, as described above, to generate an updated local model 404. When the global model 402 is updated, for example, as described below, then local model selection as shown in FIG. 4 may be repeated to generate the updated local model 404 based on the refined global model.
• an abstract model template 500 can be defined for all cohorts.
• This abstract model template 500 may be used, for example, for model generation for new cohorts, to generate original global models or the like.
• the template 500 may input the static features 104 and the semi-static features 106 into a multilayer perceptron neural network 502.
• the template 500 can further input the dynamic features 108 into a long short-term memory neural network 504.
• the output from both networks 502, 504 may provide an aggregation model 506 for multiple-label prediction.
• the output of the template 500 may be an output of multi-labels, where the template 500 can be used for all future cohort multiple-label prediction.
• a local model store 602 can include information on at least the current local model at the local site.
• the local model store 602 can also include information on other local models at other sites for a given cohort.
• the local data store 604 may be used to help identify cohorts 606 or create new cohorts 608.
• two models may be operated with data from the local data store 604 to generate a mismatch analysis 610.
• Model tuning and/or updating 612 may be performed based on the mismatch analysis 610, and the updates may be provided to a master model store 614.
• the system may provide global model updates based on one or more assessment methods.
• one method can include accessing the weights of each local model for a given cohort by averaging them (such as with a weighted average based on the number of assets in each local site) to generate an updated global model.
• the global model may be updated by accessing only the last layer’s weights of each model and averaging them to generate the global model. In this embodiment, the lower layers’ weights would be the same for each local model.
• the global model may be updated by using an ensemble approach to create a new global model based on each individual local model for a given cohort. The global model may be updated periodically or when new or updated local models are provided from one or more local sites.
• FIG. 7 presents an illustrative process 700 related to the establishment of the system 250, including local and global model generation.
• FIG. 8 presents an illustrative process 800 related to local site selection of a local model.
• Processes 700, 800 are illustrated as a collection of blocks, in a logical flowchart, which represents a sequence of operations that can be implemented in hardware, software, or a combination thereof.
• the blocks represent computer-executable instructions that, when executed by one or more processors, perform the recited operations.
• computerexecutable instructions may include routines, programs, objects, components, data structures, and the like that perform functions or implement abstract data types.
• routines programs, objects, components, data structures, and the like that perform functions or implement abstract data types.
• any number of the described blocks can be combined in any order and/or performed in parallel to implement the process.
• the process 700 for establishing a system for asset failure prediction includes an act 710 of partitioning features of the assets into static, semi-static, and dynamic features related to handling the characteristics of asset dynamics in different time scales.
• An act 720 can generate a template model to ensure the common model structure.
• FIG. 5, discussed above, provides an example of template model generation.
• An act 730 can include forming cohorts based on the static and semi-static features to group similar assets and allow a cohort specific model. In some embodiments, the system can automatically generate the cohorts based on the static and semi-static features.
• An act 740 can include training the local model with local knowledge for each cohort based on various failure types for the asset.
• An act 750 can include sharing the local knowledge to a central database and creating a global model that consolidates or aggregates the local knowledge to global knowledge for each cohort.
• the central database can distribute the cohort definitions, global model and local models to each local site.
• the process 800 can include an act 810 where a local site can select a local model from a pool of local models.
• the local site can also select the global model.
• the local site can select the best model as the new model based on performance of the two selected models.
• the local site can carry out performance and mismatch analysis of the best model.
• the local site can tune the selected model and send data back the central database.
• FIG. 9 provides a functional block diagram illustration of a computer hardware platform 900 that can be used to implement a particularly configured computing device that can host an asset failure prediction engine 950.
• the asset failure prediction engine 950 can include a cohort generation module 952, a global model database 954, a local model pool 956 and a model tuning module 958.
• the computer platform 900 may include a central processing unit (CPU) 910, a hard disk drive (HDD) 920, random access memory (RAM) and/or read only memory (ROM) 930, a keyboard 950, a mouse 960, a display 970, and a communication interface 980, which are connected to a system bus 940.
• CPU central processing unit
• HDD hard disk drive
• RAM random access memory
• ROM read only memory
• the HDD 920 has capabilities that include storing a program that can execute various processes, such as asset failure prediction engine 950, in a manner described herein.
• These computer readable program instructions may be provided to a processor of an appropriately configured computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks.
• These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, a programmable data processing apparatus, and/or other devices to function in a manner, such that the computer readable storage medium having instructions stored therein comprises an article of manufacture including instructions which implement aspects of the function/act specified in the flowchart and/or block diagram block or blocks.
• the computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other device to cause a series of operational steps to be performed on the computer, other programmable apparatus or other device to produce a computer implemented process, such that the instructions which execute on the computer, other programmable apparatus, or other device implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
• each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s).
• the functions noted in the blocks may occur out of order noted in the Figures.
• two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved.
• Relational terms such as first and second and the like may be used solely to distinguish one entity or action from another without necessarily requiring or implying any actual such relationship or order between such entities or actions.
• the terms “comprises,” “comprising,” or any other variation thereof, are intended to cover a non-exdusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
• An element proceeded by “a” or “an” does not, without further constraints, preclude the existence of additional identical elements in the process, method, article, or apparatus that comprises the element.

Landscapes

• Engineering & Computer Science (AREA)
• Theoretical Computer Science (AREA)
• Physics & Mathematics (AREA)
• Business, Economics & Management (AREA)
• General Physics & Mathematics (AREA)
• Health & Medical Sciences (AREA)
• General Health & Medical Sciences (AREA)
• General Engineering & Computer Science (AREA)
• Human Resources & Organizations (AREA)
• Economics (AREA)
• Computational Linguistics (AREA)
• Data Mining & Analysis (AREA)
• Software Systems (AREA)
• Life Sciences & Earth Sciences (AREA)
• Molecular Biology (AREA)
• Mathematical Physics (AREA)
• Evolutionary Computation (AREA)
• Biophysics (AREA)
• Biomedical Technology (AREA)
• Artificial Intelligence (AREA)
• Computing Systems (AREA)
• Marketing (AREA)
• Strategic Management (AREA)
• General Business, Economics & Management (AREA)
• Tourism & Hospitality (AREA)
• Entrepreneurship & Innovation (AREA)
• Primary Health Care (AREA)
• Public Health (AREA)
• Water Supply & Treatment (AREA)
• Operations Research (AREA)
• Quality & Reliability (AREA)
• Mechanical Engineering (AREA)
• Computer Hardware Design (AREA)
• Management, Administration, Business Operations System, And Electronic Commerce (AREA)
• Databases & Information Systems (AREA)

Priority Applications (3)

Applications Claiming Priority (2)

Publications (1)

Family

ID=81941542

Family Applications (1)

Country Status (5)

Families Citing this family (7)

Citations (7)

Family Cites Families (28)

• 2020
  • 2020-12-15 US US17/123,088 patent/US12182771B2/en active Active
• 2021
  • 2021-12-02 DE DE112021005868.1T patent/DE112021005868T5/de active Pending
  • 2021-12-02 WO PCT/IB2021/061237 patent/WO2022130098A1/en not_active Ceased
  • 2021-12-02 CN CN202180080228.4A patent/CN116601632A/zh active Pending
  • 2021-12-02 JP JP2023534934A patent/JP7724861B2/ja active Active

Patent Citations (7)

Also Published As

Similar Documents

Legal Events