EP3827387A1 - Analyse pronostique systématique avec modèle causal dynamique - Google Patents

Analyse pronostique systématique avec modèle causal dynamique

Info

Publication number
EP3827387A1
EP3827387A1 EP18779087.8A EP18779087A EP3827387A1 EP 3827387 A1 EP3827387 A1 EP 3827387A1 EP 18779087 A EP18779087 A EP 18779087A EP 3827387 A1 EP3827387 A1 EP 3827387A1
Authority
EP
European Patent Office
Prior art keywords
causal
model
topology
variable
inference
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
EP18779087.8A
Other languages
German (de)
English (en)
Inventor
Justinian Rosca
Arquimedes Martinez Canedo
Hasan Sinan BANK
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Siemens Mobility Austria GmbH
Original Assignee
Siemens Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Siemens Corp filed Critical Siemens Corp
Publication of EP3827387A1 publication Critical patent/EP3827387A1/fr
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR 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
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • 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
    • G05B17/00—Systems involving the use of models or simulators of said systems
    • G05B17/02—Systems involving the use of models or simulators of said systems electric
    • 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
    • G05B23/0218—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults
    • G05B23/0243—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model
    • G05B23/0245—Electric testing or monitoring by means of a monitoring system capable of detecting and responding to faults characterised by the fault detection method dealing with either existing or incipient faults model based detection method, e.g. first-principles knowledge model based on a qualitative model, e.g. rule based; if-then decisions
    • G05B23/0248—Causal models, e.g. fault tree; digraphs; qualitative physics

Definitions

  • This application relates to automation and control. More particularly, this application relates to prognostic analysis during runtime of an automation and control system.
  • Prognostic analysis can be applied to determine remaining lifetime of a functional component susceptible to mechanical wear, electrical degradation, and/or aging.
  • Data driven prognostics typically extrapolate to a damage threshold from historical data.
  • Model based prognostics typically estimate remaining useful life based on physical models that rely on relationships between degradation and operational conditions. For example, a model can be generated to analyze causal relationships of variables such as environmental and operational conditions measured by sensors.
  • current model driven techniques or data-driven techniques monitor elements of a system rather than the entire system, which fails to analyze systematic causal effects on a component and fails to predict a systematic outcome, such as how long until the system performance degradation reaches a failure threshold.
  • conventional prognostic techniques analyze the elements in a quantitative manner rather than a probabilistic manner, and fail to consider the changes to causal relationships over time when modeling a system.
  • Embodiments of the present invention address and overcome one or more of the above shortcomings and drawbacks, by providing methods and systems related to construction and use of prognostic models for embedded analytics and reasoning. More specifically, the prognostic models described herein stand as foundation for knowledge-based approaches in evidential reasoning about failures, root cause analysis, diagnostic and prognostic analytics. As described in further detail below, these failure models may be constructed by combining graph-based (i.e., graphical) models and statistical models with state information of operational data. The methods and systems described herein provide a technical solution for prognostic modeling in which the model automatically identifies knowledge gaps and generates requests for missing information key to better convergence. Additionally, the model topology evolves to reflect changing causal relationships and sends feedback data to a knowledge base for automatically maintaining an accurate and current knowledge base of the OE.
  • the prognostic models described herein stand as foundation for knowledge-based approaches in evidential reasoning about failures, root cause analysis, diagnostic and prognostic analytics. As described in further detail below, these failure models may be
  • a hierarchical causal model of an operating environment may be constructed using sensor data, fault detector data, prior knowledge about system variables and states, and one or more statistical descriptions of the system (e.g., cause likelihoods, fault probabilities and conditional relationships from service or prior operation in specified conditions). Additionally, the causal model may include event and variables specifying normal and abnormal system states. Conditional probabilistic reasoning may be performed for prognostic purposes on a system using the causal model to derive knowledge related to potential or actual system failures and to perform prognostic analysis.
  • a system for performing prognostic analysis of an automation system includes at least one processor and a non-transitory, computer-readable storage medium in operable communication with the at least one processor.
  • the computer- readable storage medium contains one or more programming instructions that, when executed, cause the at least one processor to receive or retrieve sensor data from a plurality of sensors associated with the automation system and to construct a causal model of the automation system using the sensor data and one or more statistical models of the automation system.
  • the instructions further cause the processor(s) to perform prognostic analytics and reasoning on the automation system using propagation of uncertainties in the causal model to derive knowledge related to potential or actual industrial system failures.
  • the technical features solved by the disclosed embodiments include: an expressive systematic model of causal relationships that cannot be understood solely through logical expressions or first principles; agile algorithms to assemble, populate, configure, change, and resolve uncertainty in an integrated causal model; and adaptive ability to integrate new expert knowledge sources during the process.
  • FIG. 1 is a block diagram of a systematic prognostic analysis engine in accordance with embodiments of the disclosure.
  • FIG. 2 is a flow diagram for an example of a process for constructing an integrated causal model in accordance with embodiments of the disclosure.
  • FIG. 3 is a flow diagram for an example of a process for analyzing the integrated causal model in accordance with embodiments of the disclosure.
  • FIG. 4 illustrates an exemplary computing environment within which embodiments of the disclosure may be implemented.
  • a physical component can be modeled as a virtual representation (i.e. , a living digital representation), or digital twin, which co-evolves with the real world.
  • a virtual representation i.e. , a living digital representation
  • digital twin which co-evolves with the real world.
  • candidates for digital twin modeling include mechanical elements of industrial systems, manufacturing systems, and railway trains.
  • an integrated causal model may be generated as a model of causal relationships in a system’s operational environment (OE) that co-evolves with the real world system.
  • the integrated causal model may analyze time series data about OE entities, behaviors, events, and actions in a formally grounded probabilistic framework.
  • the integrated causal model may be designed as a hierarchical dynamic Bayesian model to capture the entities in the OE, causal relationships, and beliefs about an OE entity’s state, to perform prognostic reasoning, where causal analysis algorithms can extract timely insights from a continuous stream of information with rich structure and connections.
  • the integrated causal model may be constructed as a continuously evolving probabilistic graphical model which inherently tracks multiple dimensions (e.g., space, time and frequency) and facilitates inference and learning during runtime. Inference with and about the integrated causal model may include runtime probabilistic reasoning to address pre-specified goals (e.g., marginal probability of a health variable) as well as uncertainty propagation.
  • the integrated causal model manages prognostics for systems in which key relationships may significantly vary over time and situation.
  • a systematic prognostic analysis (SPA) engine may generate the integrated causal model and store topology snapshots of the modeled OE over time. Past snapshots may provide a historical perspective that can be used to identify known causal patterns with supervised learning, and unknown patterns with unsupervised learning.
  • the probabilistic models may be used to predict outcomes given the current OE status and a set of proposed actions.
  • the methods and systems described herein provide a technical solution for prognostic modeling in which the model automatically identifies knowledge gaps and generates requests for missing information key to better convergence. Additionally, the model topology evolves to reflect changing causal relationships and sends feedback data to a knowledge base for automatically maintaining an accurate and current knowledge base of the OE.
  • FIG. 1 shows a block diagram of a systematic prognostic analysis engine in accordance with embodiments of this disclosure.
  • SPA engine 120 may include an interface 130, and computational tools including a causal model assembly module 140, a meta-analysis module 150 and an inference module 160.
  • the computational tools of SPA engine 120 may work in conjunction to tailor a probabilistic model, perform effective inference in large models and to support both quantitative and qualitative analysis of the OE.
  • the SPA engine 120 may define a catalogue of micro structures based on parent-child causal relationships, and structured conditional probability table algorithms to exploit topologies based on the catalogue to effectively build tractable models. Metrics may be developed to control memory use, computational performance and output quality of reference data sets.
  • the interface 130 such as an application programming interface (API) for exchanging information between SPA engine 120 and a data and knowledge base 102, may include a data and knowledge engineering interface 131 and an operation and maintenance interface 132.
  • the data and knowledge interface 131 may be used to exchange data and knowledge base data, such as OE statistical features 104 and a library of causal templates 106.
  • the data and knowledge base 102 may be initially populated with data based on expert knowledge and design engineering information of the OE.
  • new data and data updates may be fed back on data bus 108 to the data and knowledge base 102 to reflect learning of the OE based on the digital twin modeling.
  • the operation and maintenance interface 132 may be used to exchange operation and maintenance data 103 with the SPA engine 120. For example, maintenance of components in the OE may be recorded in operation and maintenance logs, and such data may be retrieved by the SPA engine 120 on data bus 105 as age related data for component models, useful for prognostic analysis.
  • the operation and maintenance interface 132 In response to operator requests for prognostic analysis and information sent on data bus 105 to the operation and maintenance interface 132 to initiate a process thread in SPA engine 120, which may access existing model data, such as model snapshots 1 15, or may initiate a new forward reasoning based inference analysis to generate a prognostic estimation and/or a predicted outcome in response to the request.
  • prognostic reports may be sent on data bus 105 for updating the operation and maintenance data base 103, including indication of OE entities needing maintenance.
  • the causal model assembly module 140 may functionally tailor a topology representation of the OE and probabilistic elements of a causal model, and may include algorithms for functions such as topology construction 141 , snapshot manager 142, and structured conditional probability table generation 143.
  • the causal model assembly module 140 may generate causal models, which may be stored as a series of topology snapshots to create an integrated causal model 1 10, which captures temporal dynamics with the various snapshots.
  • a causal model may be based on a graphical model, such as a map of entities and their causal relationships retrieved from the causal template library 106, which describes variables in the domain and their possible causal relationships.
  • the causal template library 106 may store sets of event and normality variables that describe concepts in a domain, such as a prognostic domain. Random variables for the model, given by a set V, may affect the failure status and the causal relationships amongst them, given by set £. Domain events and normality variables may be subset of V.
  • the domain for each variable V may be specified (e.g., categorical, such as Boolean, or n-valued variable; continuous, real-valued variables).
  • a computational or statistical procedure or model for computing the value of each of the event and normality variables may be provided.
  • a statistical procedure could be, for example, a one-class support vector machine that indicates if the value of the variable is normal, or a classification procedure that classifies the incoming data (typically time series data from various sensor inputs).
  • the OE statistical features 104 may include time series data acquired, for example, from sensors present in event detectors, classifiers, and detectors.
  • the knowledge base 102 such as the likelihood of specific states of time and the order of fault causes over time, may be generated by SPA engine 120 analysis and stored as the OE statistical features 104.
  • the integrated causal model 1 10 may include a series of topologies representing an evolution of the causal model over time. As shown in FIG. 1 , a portion of the series of topologies includes topology snapshots 1 1 1 1 , 1 12, 1 13, 1 14 in the time domain for time ti, ti+1 , tj, tk. The integrated causal model 1 10 is expected to remain stable for substantial time periods between snapshots 1 1 1 , 1 12, 1 13, 1 14, where a snapshot is triggered by updates made to the topology.
  • a causal model may be formalized by directed graphs G(V,E), For example, snapshot 1 1 1 includes variables A, C, D and B of set V, linked by edges of set E.
  • variables may include random or stochastic variables such as an alarm state, a sensor’s active state, an entity A that triggers a specific reading in sensor B, weather conditions, a faulty sensor, and the like.
  • Some variables may be hidden variables, not directly observable or sensor related, but influencing the OE nonetheless.
  • the directed edges indicate causal relationships between the variables. For example, in snapshot 1 12, variables C and D are causally influenced by variable A as shown by directed edges. Edge weights may be calculated based on pairwise causality measurements.
  • Variables may be concepts that are internal to the OE (e.g., sensor value regimes corresponding to mechanical components, state information, alarm state, and the like) or external (e.g.
  • the random variables V may be based on data and knowledge base information 102, such as OE statistical features 104, and stored information 103 obtained from design, operation, and maintenance engineering processes.
  • the topology construction algorithms 141 may compile a hierarchical dynamic Bayesian model during model assembly and model learning, which represents the integrated causal model 1 10.
  • the causal model may be representative of a system in part or in entirety, and constructed using a hierarchy of causal variables related to the OE, including classes and subclasses related by an “is-a” relationship for example.
  • the failure model may be further constructed using an ontology of failure variables related to the system comprising classes and subclasses related by a“has-an-effect-on” relationship or an“influences” relationship.
  • Analysis of the integrated causal model 1 10 may include a trace of a time series of streams of statistical features acquired from OE data and knowledge base 104, which may then be subjected to prognostic analysis. The structure and the changes of the trace over time may be used by supervised/unsupervised machine learning algorithms to extract known and unknown causal relationships, and trace back in time the root causes of observed behaviors.
  • the causal relationships in topology shapshots may evolve over time, such as an edge weight value change to edge 1 15 from snapshot 1 1 1 to snapshot 1 12.
  • Another example of causal relationship updates is shown by edge 1 15 in snapshot 1 12 becoming edge 1 16 in snapshot 1 13, illustrating a change in causal relationship for variables A, B and D.
  • Key causal relationships may vary significantly over time and specific situation, which is captured by the integrated causal modeling described herein.
  • the topology construction algorithms 141 may manage variants of the topology. Learning by the dynamic Bayesian modeling performed by causal model assembly module 141 may determine that previous causal relationships are no longer valid to explain the systematic statistical features of the OE.
  • topology snapshots 1 13 and 1 14 For example, evolving topology models are represented in topology snapshots 1 13 and 1 14 in which variables X and Y are introduced in topology snapshot 1 14, while variable A has been removed.
  • the dynamic nature of the Bayesian model is evident by the time evolution of causal relationships, with a present state influenced by previous states, and the recognition that the conditional probabilities for the causal relationships can be strengthened by observed data.
  • the snapshot manager algorithm 142 may be triggered automatically by meta-analysis or configured to optimize the periodicity of snapshot captures as the topology evolves. For example, when infrequent changes occur to the topology, redundancy of snapshots can be minimized by increasing the period between snapshot capture and storage 1 17. Should topology changes become more rapid, the snapshot manager 142 may shorten the snapshot capture period. By doing so, the granularity of data collection may be controlled. Management of the snapshots may also include purging of redundant snapshots to improve efficiency of scanning snapshots for forward and backward reasoning by inference module 160. The causal model efficiency may also be controlled. As an example, if inference is more complex than typical because of missing data streams, then it may be beneficial to revisit previous snapshots in order to make topology changes to simplify the model in order to gain speed and recover memory resources.
  • the structured conditional probability table module 143 may include algorithms for generating conditional probability tables using pre-defined specialized heuristic rules within a dynamic Bayesian model.
  • a captured local structure may affect the algorithms with respect to choice of model and inference, as well as the size of generated tables based on number of table components. As the number of components captured increases, more structure representation may be induced. Representation may be over time (e.g., a random variable from a previous time step), or at present time. Structured representations require larger conditional probability tables that are constructed semi-automatically using rules or mathematical formulas. Elements of causal failure modeling may be defined automatically or manually during a“learning” or acquisition phase of the model.
  • An initialization may use conditional probability tables for all variables in V given their parents (other nodes in V), according to the structure of graph G. Each of these constitutes the graphical model for reasoning with uncertainty, and it may be loaded at run time initialization.
  • Failure and maintenance data, and engineering knowledge from the domain e.g., in the form of simulators using physics- based models
  • provides numerical information such as the severity of a failure, frequency of occurrence of a failure, the likelihood of accident based on the failure, the tolerability of failure errors, and statistics regarding the deviations of fault detector measurements from normal, etc.
  • These quantitative descriptions extracted as statistics from service and maintenance records data 103 can be automatically loaded to define priors and conditional probabilities needed for initialization. Where these are missing, uniform likelihoods or domain expert input may be used.
  • Bayesian networks operating on directed acyclic graphs (typically), or more generally undirected graphs. Knowledge may be compiled automatically (learned and formalized) into a graphical model.
  • Traditional methods can learn the structure of a Bayesian network by a randomized search provided a fitness or quality function is given to evaluate its topology in terms of capacity to explain the evidence from the domain.
  • Examples of Bayesian network structure include, but are not limited to node types and functional dependence referred to as noisy, noisy-or, noisy-and, noisy-max. Parts of the causal network can be extracted from a domain ontology.
  • a conditional probability table may be generated for each of the variables A, B, C and D, such as the probability P for variable C given A and D is represented by P(C
  • the table may include P values for each possible value combination of A and D.
  • P values can be learned or defined by a model with heuristic rules (e.g., max rule) using priors computed according to the Bayesian model based on the knowledge base. For example, an entity of the OE may have a known failure rate of 0.005 based on evidence, and from this prior knowledge, the Bayesian model can compute a posterior probability grounded by the priors.
  • the conditional probability table may be attached to each variable node in the topology graph, and may include a time label for the purpose of tracking the time dependencies of the graph evolution over time.
  • Various discrete, continuous and mixed distributions including but not limited to Gaussian, Poisson, Boltzmann, and Laplace, may be applied by the conditional probability table module 143 to derive probability tables for discrete and continuous variables.
  • the meta-analysis module 150 may include reasoning algorithms to analyze the input data streams, such as data and knowledge base data on data bus 108, and requests for prognosis information on data bus 105.
  • the meta-analysis module 150 may include a knowledge gap detection algorithm 152 that can analyze the topology snapshots of the integrated causal model 1 10 to identify gaps in the knowledge which impede inferences for prognosis. Such gaps can be conveyed back to the user in the form of a request for additional knowledge and data to improve the integrated causal model 110.
  • the meta-analysis module 150 may include a topology optimization algorithm 151 which may operate as a daemon running in the background of the SPA engine 120 to monitor the topology and dynamically modify the models over time.
  • the topology optimization algorithm 151 may determine that the complexity for deriving a particular inference by the integrated causal model 1 10 may exceed preset limitations for computer resources, such as a threshold on memory utilization or duration of computation. For example, several factors of the stored snapshots 1 17 may be examined, including a count of parent-child relationships of a causal solution, or a count of influence paths and loops from any given node to another node, and based on these factors, a graph of theoretical properties can be determined as a gauge of complexity for the inference solution.
  • the topology optimization algorithm 151 may eliminate one or more nodes and/or edges from the topology in historical snapshots to improve the time efficiency in response to a determination that the complexity exceeds a threshold.
  • the topology optimization algorithm 151 may adjust probabilities in the condition probability tables, which may occur with operator assistance, using historical trace of the model (historical snapshots). With a modified topology evolution, the optimization may reanalyze the complexity and make further adjustments to resolve and maintain the complexity of the integrated causal model 1 10.
  • the topology optimization algorithm 151 provides an advantage over static probability models of conventional methods that can only make adjustments in a forward direction as a reactive measure in a diagnostic domain.
  • the meta-analysis module 150 of SPA engine 120 allows for backward and forward optimization of a dynamic Bayesian model.
  • the meta-analysis module 150 may include a system of system (SoS) sensitivity algorithm 153 which executes to locate variables within the integrated causal model having causal influence on target variables.
  • SoS sensitivity algorithm 153 may identify a target variable, which may be related to a prognosis request, and locate the target variable within the topology snapshots. Using the causal relationships of the target variable as a map, the influencing variables may be isolated and incremental adjustments may be applied to one or more influencing variables.
  • a target probability may be in the range (0.9-1 ) for a target variable, while the obtained value in the conditional probability table is 0.8.
  • the SoS sensitivity algorithm 153 may reassess values of influencing variables, such as hidden variables, and modify the variable by an incremental adjustment.
  • the target variable probability value may be monitored for its reaction sensitivity to the adjustment further adjustments may be made until convergence is achieved.
  • the SoS sensitivity algorithm 153 may also be configured to determine the degree of influence by the influencing variables from strongest to weakest influence, and adjust influencing variables accordingly. Hence, the target probability values can be maintained in the conditional probability tables for particular variables.
  • the inference module 160 may be used to perform forward and backward reasoning.
  • an uncertainty propagation algorithm 161 may execute in conjunction with forward reasoning algorithm 162 and backward reasoning algorithm 163 to generate forward traces of snapshots and past traces of snapshots based on the probabilistic models. For example, if some variables are based on assumed values, such as in the case of hidden variables, backward propagation settings by the uncertainty propagation algorithm 161 may be determined. Using the backward propagation, the backward reasoning algorithm 163 may execute backward reasoning with adjustment to one or more variables which were set based on choices that now appear incorrect, which may provide a more accurate model of the present state of the OE.
  • the inference module 160 may adjust related causal variables in a backward propagation to derive a more accurate P value for variable A.
  • a forward trace may be generated by the forward reasoning algorithm 162 by applying a forward propagation. Accordingly, marginal probability distributions may be applied to achieve inference functionality based on existing evidence.
  • the uncertainty propagation algorithm 161 may serve a useful function to solve a prognostic solution to a faulty sensor represented by variable A. For example, over a series of snapshot instances, no sensor readings for the variable A have been received, and any assumed values for variable A would bias the topology. Without other reading to rely on, the variable A cannot be grounded. At a later time, the sensor is replaced, and readings are received from OE statistics data 104. Using a backward propagation, a backward inference may be derived using the new knowledge for variable A, which may affect causally related variables to improve the accuracy of the causal model.
  • pre-processing filtering techniques may be used to check the sanity of the input data before feeding it to the machine learning algorithms of the SPA engine 120.
  • the SPA engine 120 may generate requests to an operator for additional data discovered as knowledge/data gaps by the knowledge gap detection algorithm 152.
  • knowledge gap detection algorithm 152 may find a mismatch in a prediction of snapshots, such as where a previous time step snapshot 1 12 is compared to a current time step snapshot 1 13.
  • the operator may be consulted for clarification by sending a knowledge request message on data bus 105.
  • the interface 130 may specify the format and functionality of feature streams on data buses 105, 108.
  • APIs of the data and knowledge interface 131 may specify formats of features of knowledge objects, statistical priors about causal dependencies, and the variables.
  • the APIs may also specify format of sensor/feature and parameter streams (e.g., environment parameters and time series of features), and adaptation streams (output of the ICM meta-analysis algorithms for ICM updates), and snapshot management functions.
  • APIs of the operation and maintenance interface 132 may specify format of human interface interaction streams, such as query-answer streams with user input and prognostic inference reports.
  • the dynamic Bayesian model approach of the disclosed embodiments permits reasoning under uncertainty.
  • the SPA engine 120 may employ strategies depending on the information that is available from data and knowledge base 102.
  • a dynamic Bayesian Network may operate with optimizations, such as micro structures and algorithms operating on structured representations rather than directly on conditional probability tables in order to efficiently represent conditional probabilities.
  • Other optimizations may be inferences at a higher specificity level and efficient inference algorithms (e.g., forward reasoning algorithm 162 and backward reasoning algorithm 163) operating on low tree width topologies to deal with large scale networks.
  • a hierarchical dynamic Bayesian network may be optimized as a dynamic oriented object Bayesian network, which may represent the domain as a hierarchy of interconnected objects and which may have evolved as a paradigm to succinctly model complex interactions. This optimization may model complex and evolving OEs.
  • FIG. 2 is a flow diagram for an example of a process for constructing an integrated causal model in accordance with embodiments of this disclosure.
  • time series data with OE statistical features and/or causal templates may be fed to the SPA engine 120 for construction of the integrated causal model, where the data is used for learning (e.g., conditional probability tables and evolving causal relationships).
  • a causal model topology may be constructed according to a hierarchical dynamic Bayesian model with nodes representing random variables and edges representing directional causal relationships between the nodes. Structured conditional probability tables may be produced at 204 for variables of the causal model and attached to nodes of the graphical topology.
  • Efficiency gains in conditional probability table generation may be achieved through operator support by applying filters to reduce the quantity of conditional probability table information.
  • algorithmic rules may be applied by the structured conditional probability table module 143.
  • the topology may be optimized to ensure meta-analysis using inference rules on the Bayesian causal model can be performed within computational resource constraints. For example, during the learning phase of the causal model, the time series data inputs 201 may be labeled, and the outputs of the causal model may be evaluated for error to test whether inferences by Bayesian causal model are accurate.
  • the topology optimization may perform a tradeoff analysis between high accuracy with high model complexity and lower acceptable accuracy at lower resource consumption by reducing the model complexity.
  • topology optimization module 151 may be executed by the topology optimization module 151 as necessary to achieve the optimization.
  • knowledge gaps may be detected by meta-analysis performed on input data and/or performed on the optimized topology to identify information needed for tuning causal solutions as part of causal model learning.
  • missing knowledge or data identified by the knowledge gap detection may be packaged as a request message to a system operator at 208.
  • the process may repeat from 201 , where missing knowledge information is provided to the SPA engine 120 so that the causal model topology may be updated at 203, any new conditional probability tables needed may be generated at 204, and the topology optimization 205 may be repeated if necessary. If at 207, no knowledge gaps are detected, the constructed integrated causal model is now available at 209 for runtime prognostic analysis.
  • FIG. 3 is a flow diagram for an example of a process for prognostic analysis using the integrated causal model in accordance with embodiments of the disclosure.
  • an operator may submit a prognostic request 301 using a graphical user interface (GUI), to seek a prediction for failure of a system component (e.g., such as an axle in a railway car), six months from now.
  • GUI graphical user interface
  • the form of the request may be plain text or from a menu selection that offers various system components for which a failure prognosis can be requested.
  • the causal model has been developed for the domain of prognostic analysis of components in a railway car, meaning an ontology for the random variables relates to causal relationships to failure events of the various components.
  • APIs of the interfaces 130 are programmed to enable the operator to communicate with the SPA engine 120 in terminology corresponding to the particular system.
  • the causal model 1 10 having been learned by the SPA engine 120 can locate the variables causally related to the OE of the axle, such as age of springs, vibration sources, sensor readings, maintenance records, events from event logs (e.g., axle failure data in similar railway cars), and the like.
  • SPA engine 120 may analyze the integrated causal model at 302 in response to the prognosis request.
  • a knowledge gap detection 303 may be performed to check for missing data, and if it is determined that no additional data is needed, the integrated causal model may be analyzed to generate a prognostic report 309.
  • Analysis of the integrated causal model at 302 may include generation of a forward reasoning trace of the integrated causal model by the inference module 160 at 308, from which a prognostic report 309 is generated and transmitted to the operator at the GUI, indicating a probability of failure of the axle in six month’s time. If the result is unsatisfactory, such as a strong likelihood of failure, another prognostic request 301 may be submitted as a what-if analysis (i.e., as an alternative scenario prognosis), having a candidate maintenance event suggested as virtual evidence (e.g., replacement of a bushing in the axle assembly).
  • the reasoning trace by inference module 160 may include alternate variables in the topology of the causal model to reflect a new bushing and in general a change of state of the system, and a new probability of failure in prognostic report 309 may be generated using the integrated causal model.
  • the prognostic report 309 may include suggested remedies to implement in the automation system in order to prolong the life of the component, and defer or prevent the predicted failure.
  • prognostic analysis at 302 and inference processing at 308 is determined by SPA engine 120 to be taking too long or if memory is depleted, an optimization of the topology may be triggered at 307, which may eliminate some causal model nodes or reduce size of one or more conditional probability tables.
  • a missing data/knowledge request 304 may be generated and transmitted to the operator.
  • time series data with OE statistical features and/or causal templates 306 may be sent to the SPA engine 120 for an update of the causal model topology at 305.
  • the OE statistical features of a knowledge base may be updated with feedback of knowledge updates at 310, which is beneficial to maintaining a current and accurate knowledge base for the OE entities.
  • the SPA engine 120 may periodically perform topology optimization 307 to assess conditional probability tables and determine through machine learning which variables and causal relationships require updates, by addition and/or removal of nodes and edges in the topology graphs.
  • the knowledge updates 310 may be fed back to the knowledge base for maintaining a current and accurate knowledge base for the OE entities.
  • the technical solution in this implementation illustrated in FIG. 3 is a system configured to identify potential failures having a strong likelihood that may be influenced by hidden, undetected variables learned from prognostic analysis of a dynamic Bayesian model, in which evolving causal relationships are tracked to update the model for accurately reflecting current states.
  • the methods and systems described herein greatly simplify construction of an effective prognostic network using automatic causal model assembly with meta- analysis, which is applicable to a wide range of situations requiring embedded edge intelligence. While the system and corresponding OE have been described with respect to a prognostic domain related to an automation and control system, the methods and systems as disclosed may also be implemented in other forms of prognostic analysis, such as causal events tied to social and political events (e.g., prognostic analysis of crowd riots), and design choices in product lifetime cycle, which could be modeled with nodes in a causal model graph and analyzed in accordance with embodiments of the disclosure.
  • FIG. 4 illustrates an exemplary computing environment 400 within which embodiments of the disclosure may be implemented.
  • the computer system 410 may include a communication mechanism such as a system bus 421 or other communication mechanism for communicating information within the computer system 410.
  • the computer system 410 further includes one or more processors 420 coupled with the system bus 421 for processing the information.
  • the processors 420 may include one or more central processing units (CPUs), graphical processing units (GPUs), or any other processor known in the art. More generally, a processor as described herein is a device for executing machine- readable instructions stored on a computer readable medium, for performing tasks and may comprise any one or combination of, hardware and firmware. A processor may also comprise memory storing machine-readable instructions executable for performing tasks. A processor acts upon information by manipulating, analyzing, modifying, converting or transmitting information for use by an executable procedure or an information device, and/or by routing the information to an output device.
  • CPUs central processing units
  • GPUs graphical processing units
  • a processor may use or comprise the capabilities of a computer, controller or microprocessor, for example, and be conditioned using executable instructions to perform special purpose functions not performed by a general purpose computer.
  • a processor may include any type of suitable processing unit including, but not limited to, a central processing unit, a microprocessor, a Reduced Instruction Set Computer (RISC) microprocessor, a Complex Instruction Set Computer (CISC) microprocessor, a microcontroller, an Application Specific Integrated Circuit (ASIC), a Field-Programmable Gate Array (FPGA), a System-on-a-Chip (SoC), a digital signal processor (DSP), and so forth.
  • RISC Reduced Instruction Set Computer
  • CISC Complex Instruction Set Computer
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • SoC System-on-a-Chip
  • DSP digital signal processor
  • processor(s) 420 may have any suitable microarchitecture design that includes any number of constituent components such as, for example, registers, multiplexers, arithmetic logic units, cache controllers for controlling read/write operations to cache memory, branch predictors, or the like.
  • the microarchitecture design of the processor may be capable of supporting any of a variety of instruction sets.
  • a processor may be coupled (electrically and/or as comprising executable components) with any other processor enabling interaction and/or communication there-between.
  • a user interface processor or generator is a known element comprising electronic circuitry or software or a combination of both for generating display images or portions thereof.
  • a user interface comprises one or more display images enabling user interaction with a processor or other device.
  • the system bus 421 may include at least one of a system bus, a memory bus, an address bus, or a message bus, and may permit exchange of information (e.g., data (including computer-executable code), signaling, etc.) between various components of the computer system 410.
  • the system bus 421 may include, without limitation, a memory bus or a memory controller, a peripheral bus, an accelerated graphics port, and so forth.
  • the system bus 421 may be associated with any suitable bus architecture including, without limitation, an Industry Standard Architecture (ISA), a Micro Channel Architecture (MCA), an Enhanced ISA (EISA), a Video Electronics Standards Association (VESA) architecture, an Accelerated Graphics Port (AGP) architecture, a Peripheral Component Interconnects (PCI) architecture, a PCI-Express architecture, a Personal Computer Memory Card International Association (PCMCIA) architecture, a Universal Serial Bus (USB) architecture, and so forth.
  • ISA Industry Standard Architecture
  • MCA Micro Channel Architecture
  • EISA Enhanced ISA
  • VESA Video Electronics Standards Association
  • AGP Accelerated Graphics Port
  • PCI Peripheral Component Interconnects
  • PCMCIA Personal Computer Memory Card International Association
  • USB Universal Serial Bus
  • the computer system 410 may also include a system memory 430 coupled to the system bus 421 for storing information and instructions to be executed by processors 420.
  • the system memory 430 may include computer readable storage media in the form of volatile and/or nonvolatile memory, such as read only memory (ROM) 431 and/or random access memory (RAM) 432.
  • the RAM 432 may include other dynamic storage device(s) (e.g., dynamic RAM, static RAM, and synchronous DRAM).
  • the ROM 431 may include other static storage device(s) (e.g., programmable ROM, erasable PROM, and electrically erasable PROM).
  • system memory 430 may be used for storing temporary variables or other intermediate information during the execution of instructions by the processors 420.
  • a basic input/output system 433 (BIOS) containing the basic routines that help to transfer information between elements within computer system 410, such as during start-up, may be stored in the ROM 431.
  • RAM 432 may contain data and/or program modules that are immediately accessible to and/or presently being operated on by the processors 420.
  • System memory 430 may additionally include, for example, operating system 434, application programs 435, and other program modules 436.
  • the operating system 434 may be loaded into the memory 430 and may provide an interface between other application software executing on the computer system 410 and hardware resources of the computer system 410. More specifically, the operating system 434 may include a set of computer-executable instructions for managing hardware resources of the computer system 410 and for providing common services to other application programs (e.g., managing memory allocation among various application programs). In certain example embodiments, the operating system 434 may control execution of one or more of the program modules depicted as being stored in the data storage 440.
  • the operating system 434 may include any operating system now known or which may be developed in the future including, but not limited to, any server operating system, any mainframe operating system, or any other proprietary or non-proprietary operating system.
  • the application programs 435 may include a set of computer-executable instructions for performing the causal modeling construction and analysis processes in accordance with embodiments of the disclosure.
  • the computer system 410 may also include a disk/media controller 443 coupled to the system bus 421 to control one or more storage devices for storing information and instructions, such as a magnetic hard disk 441 and/or a removable media drive 442 (e.g., floppy disk drive, compact disc drive, tape drive, flash drive, and/or solid state drive).
  • Storage devices 440 may be added to the computer system 410 using an appropriate device interface (e.g., a small computer system interface (SCSI), integrated device electronics (IDE), Universal Serial Bus (USB), or FireWire).
  • Storage devices 441 , 442 may be external to the computer system 410, and may be used to store processing data in accordance with the embodiments of the disclosure.
  • the computer system 410 may also include a display controller 465 coupled to the system bus 421 to control a display or monitor 466, such as a cathode ray tube (CRT) or liquid crystal display (LCD), for displaying information to a computer user, such as a system operator as described herein.
  • the computer system includes a user input interface 460 and one or more input devices, such as a user terminal 461 , which may include a keyboard, touchscreen, tablet and/or a pointing device, for interacting with a computer user (e.g., such as an operator as described herein) and providing information to the processors 420.
  • the display 466 may provide a touch screen interface which allows input to supplement or replace the communication of direction information and command selections by the user terminal device 461.
  • the computer system 410 may perform a portion or all of the processing steps of embodiments of the invention in response to the processors 420 executing one or more sequences of one or more instructions contained in a memory, such as the system memory 430. Such instructions may be read into the system memory 430 from another computer readable medium, such as the magnetic hard disk 441 or the removable media drive 442.
  • the magnetic hard disk 441 may contain one or more data stores and data files used by embodiments of the present invention.
  • the data store may include, but are not limited to, databases (e.g., relational, object-oriented, etc.), file systems, flat files, distributed data stores in which data is stored on more than one node of a computer network, peer-to-peer network data stores, or the like. Data store contents and data files may be encrypted to improve security.
  • the processors 420 may also be employed in a multi-processing arrangement to execute the one or more sequences of instructions contained in system memory 430.
  • hard-wired circuitry may be used in place of or in combination with software instructions. Thus, embodiments are not limited to any specific combination of hardware circuitry and software.
  • the computer system 410 may include at least one computer readable medium or memory for holding instructions programmed according to embodiments of the invention and for containing data structures, tables, records, or other data described herein.
  • the term “computer readable medium” as used herein refers to any medium that participates in providing instructions to the processors 420 for execution.
  • a computer readable medium may take many forms including, but not limited to, non-transitory, non-volatile media, volatile media, and transmission media.
  • Non-limiting examples of non-volatile media include optical disks, solid state drives, magnetic disks, and magneto-optical disks, such as magnetic hard disk 441 or removable media drive 442.
  • Non-limiting examples of volatile media include dynamic memory, such as system memory 430.
  • Non-limiting examples of transmission media include coaxial cables, copper wire, and fiber optics, including the wires that make up the system bus 421. Transmission media may also take the form of acoustic or light waves, such as those generated during radio wave and infrared data communications.
  • Computer readable medium instructions for carrying out operations of the present disclosure may be assembler instructions, instruction-set-architecture (ISA) instructions, machine instructions, machine dependent instructions, microcode, firmware instructions, state-setting data, or either source code or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, C++ or the like, and conventional procedural programming languages, such as the "C" programming language or similar programming languages.
  • the computer readable program instructions may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server.
  • the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or the connection may be made to an external computer (for example, through the Internet using an Internet Service Provider).
  • LAN local area network
  • WAN wide area network
  • Internet Service Provider for example, AT&T, MCI, Sprint, EarthLink, MSN, GTE, etc.
  • electronic circuitry including, for example, programmable logic circuitry, field-programmable gate arrays (FPGA), or programmable logic arrays (PLA) may execute the computer readable program instructions by utilizing state information of the computer readable program instructions to personalize the electronic circuitry, in order to perform aspects of the present disclosure.
  • FPGA field-programmable gate arrays
  • PLA programmable logic arrays
  • the computing environment 400 may further include the computer system 410 operating in a networked environment using logical connections to one or more remote computers, such as remote computing device 480, remote storage devices 491 (e.g., for storing the knowledge base information), and entities of the operating environment (OE) 490, such as sensors and other devices that generate information corresponding to the variables for the causal model.
  • the network interface 470 may enable communication, for example, with other remote devices 480 or systems and/or remote storage devices 491 via the network 471.
  • Remote computing device 480 may be a personal computer (laptop or desktop), a mobile device, a server, a router, a network PC, a peer device or other common network node, and typically includes many or all of the elements described above relative to computer system 410.
  • computer system 410 may include modem 472 for establishing communications over a network 471 , such as the Internet. Modem 472 may be connected to system bus 421 via user network interface 470, or via another appropriate mechanism.
  • Network 471 may be any network or system generally known in the art, including the Internet, an intranet, a local area network (LAN), a wide area network (WAN), a metropolitan area network (MAN), a direct connection or series of connections, a cellular telephone network, or any other network or medium capable of facilitating communication between computer system 410 and other computers (e.g., remote computing device 480).
  • the network 471 may be wired, wireless or a combination thereof. Wired connections may be implemented using Ethernet, Universal Serial Bus (USB), RJ-6, or any other wired connection generally known in the art.
  • Wireless connections may be implemented using Wi-Fi, WiMAX, and Bluetooth, infrared, cellular networks, satellite or any other wireless connection methodology generally known in the art. Additionally, several networks may work alone or in communication with each other to facilitate communication in the network 471.
  • program modules, applications, computer- executable instructions, code, or the like depicted in FIG. 4 as being stored in the system memory 430 are merely illustrative and not exhaustive and that processing described as being supported by any particular module may alternatively be distributed across multiple modules or performed by a different module.
  • various program module(s), script(s), plug-in(s), Application Programming Interface(s) (API(s)), or any other suitable computer-executable code hosted locally on the computer system 410, the remote device 480, and/or hosted on other computing device(s) accessible via one or more of the network(s) 471 may be provided to support functionality provided by the program modules, applications, or computer-executable code depicted in FIG.
  • functionality may be modularized differently such that processing described as being supported collectively by the collection of program modules depicted in FIG. 4 may be performed by a fewer or greater number of modules, or functionality described as being supported by any particular module may be supported, at least in part, by another module.
  • program modules that support the functionality described herein may form part of one or more applications executable across any number of systems or devices in accordance with any suitable computing model such as, for example, a client-server model, a peer- to-peer model, and so forth.
  • any of the functionality described as being supported by any of the program modules depicted in FIG. 4 may be implemented, at least partially, in hardware and/or firmware across any number of devices.
  • An executable application comprises code or machine readable instructions for conditioning the processor to implement predetermined functions, such as those of an operating system, a context data acquisition system or other information processing system, for example, in response to user command or input.
  • An executable procedure is a segment of code or machine readable instruction, sub-routine, or other distinct section of code or portion of an executable application for performing one or more particular processes. These processes may include receiving input data and/or parameters, performing operations on received input data and/or performing functions in response to received input parameters, and providing resulting output data and/or parameters.
  • a graphical user interface comprises one or more display images, generated by a display processor and enabling user interaction with a processor or other device and associated data acquisition and processing functions.
  • the GUI also includes an executable procedure or executable application.
  • the executable procedure or executable application conditions the display processor to generate signals representing the GUI display images. These signals are supplied to a display device which displays the image for viewing by the user.
  • the processor under control of an executable procedure or executable application, manipulates the GUI display images in response to signals received from the input devices. In this way, the user may interact with the display image using the input devices, enabling user interaction with the processor or other device.
  • the prognostic analysis request messages by a system operator and the knowledge gap request messages received by a system operator may be generated using a GUI.
  • the functions and process steps herein may be performed automatically or wholly or partially in response to user command.
  • An activity (including a step) performed automatically is performed in response to one or more executable instructions or device operation without user direct initiation of the activity.

Abstract

L'invention concerne un procédé et un système destinés à une analyse pronostique systématique qui comprennent un moteur d'analyse pronostique systématique comprenant des algorithmes d'assemblage de modèles causaux pour recevoir des flux de données d'éléments statistiques des entités OE sur la base de données historiques, pour construire un modèle causal intégré graphique ayant un ensemble de topologies, chaque topologie ayant des noeuds qui représentent des variables stochastiques associée aux entités OE. Pour chaque variable, des tableaux de probabilité conditionnelle structurés sont construits sur la base d'un modèle bayésien dynamique et des tableaux de probabilité conditionnelle correspondants sont associés à chaque variable. Le moteur d'analyse pronostique systématique peut comprendre des algorithmes de méta-analyse pour détecter des espaces de connaissances dans les données d'entrée utilisées pour construire les tableaux de probabilité conditionnelle structurés et optimiser la topologie sur la base des contraintes de ressources informatiques.
EP18779087.8A 2018-08-27 2018-08-27 Analyse pronostique systématique avec modèle causal dynamique Pending EP3827387A1 (fr)

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/US2018/048049 WO2020046261A1 (fr) 2018-08-27 2018-08-27 Analyse pronostique systématique avec modèle causal dynamique

Publications (1)

Publication Number Publication Date
EP3827387A1 true EP3827387A1 (fr) 2021-06-02

Family

ID=63686071

Family Applications (1)

Application Number Title Priority Date Filing Date
EP18779087.8A Pending EP3827387A1 (fr) 2018-08-27 2018-08-27 Analyse pronostique systématique avec modèle causal dynamique

Country Status (2)

Country Link
EP (1) EP3827387A1 (fr)
WO (1) WO2020046261A1 (fr)

Families Citing this family (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP4128070A1 (fr) * 2020-03-31 2023-02-08 Abb Schweiz Ag Procédé d'apprentissage machine hiérarchique pour un système d'apprentissage machine industrielle
CN111724126B (zh) * 2020-06-12 2024-03-12 北京科技大学顺德研究生院 一种工艺流程质量异常精准追溯方法及系统
CN112147974B (zh) * 2020-07-22 2022-07-12 中国石油大学(华东) 基于化工过程知识自动化的报警根原因诊断方法
DE102020120027A1 (de) 2020-07-29 2022-02-03 Liebherr-Verzahntechnik Gmbh Verfahren zur automatischen Bestimmung von Konstruktionsparametern eines Greifers
CN114080609A (zh) * 2020-11-18 2022-02-22 阿里巴巴集团控股有限公司 基于编码知识的非线性因果建模
CN112506995A (zh) * 2020-12-08 2021-03-16 青岛欧赛斯环境与安全技术有限责任公司 一种基于时间关联溯因推理的化工过程报警根源诊断方法
CN113822431B (zh) * 2021-03-19 2024-04-16 京东科技控股股份有限公司 贝叶斯网络的结构学习方法、装置及电子设备
CN113328881B (zh) * 2021-05-26 2022-05-03 南京航空航天大学 一种面向非合作无线网络的拓扑感知方法及装置、系统
WO2023003545A1 (fr) * 2021-07-21 2023-01-26 Siemens Aktiengesellschaft Contrôle de processus de production piloté par des graphiques
CN113657599A (zh) * 2021-08-20 2021-11-16 北京航空航天大学 事故因果推理方法、装置、电子设备及可读存储介质
CN113923099B (zh) * 2021-09-03 2022-12-27 华为技术有限公司 一种通信网络故障的根因定位方法及相关设备
CN114844784B (zh) * 2022-04-11 2023-08-08 桂林电子科技大学 一种可重构扫描网络的拓扑建模方法
EP4286970A1 (fr) * 2022-05-31 2023-12-06 Siemens Aktiengesellschaft Procédé et système de détection d'anomalies de capteur
CN114943304B (zh) * 2022-06-17 2023-04-07 广东工业大学 一种基于贝叶斯的3c电子产品装配误差传递方法
CN116484428B (zh) * 2023-04-27 2024-02-02 湖南工商大学 数据安全性检测系统、方法、装置及相关设备
CN117785565B (zh) * 2024-02-23 2024-05-03 深圳市创世飞达科技有限公司 一种固态存储硬盘数据恢复方法及固态硬盘

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7797062B2 (en) * 2001-08-10 2010-09-14 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US20090204237A1 (en) * 2001-08-10 2009-08-13 Rockwell Automation Technologies, Inc. System and method for dynamic multi-objective optimization of machine selection, integration and utilization
US7254514B2 (en) * 2005-05-12 2007-08-07 General Electric Company Method and system for predicting remaining life for motors featuring on-line insulation condition monitor
MX2016006764A (es) * 2013-11-27 2016-12-16 Adept Ai Systems Inc Metodo y sistema para controlar procesos dinamicos con base en un modelo artificialmente inteligente usando agentes probabilisticos.
CN110462644A (zh) * 2017-01-24 2019-11-15 西门子股份公司 用于系统的自动化和控制的认知工程技术的系统和方法

Also Published As

Publication number Publication date
WO2020046261A1 (fr) 2020-03-05

Similar Documents

Publication Publication Date Title
EP3827387A1 (fr) Analyse pronostique systématique avec modèle causal dynamique
US10733536B2 (en) Population-based learning with deep belief networks
US10192170B2 (en) System and methods for automated plant asset failure detection
US11720821B2 (en) Automated and customized post-production release review of a model
US10410135B2 (en) Systems and/or methods for dynamic anomaly detection in machine sensor data
US11288577B2 (en) Deep long short term memory network for estimation of remaining useful life of the components
US9727441B2 (en) Generating dependency graphs for analyzing program behavior
US10635094B2 (en) Failure models for embedded analytics and diagnostic/prognostic reasoning
EP3183622B1 (fr) Apprentissage basé sur une population et incluant des réseaux de croyances profondes
Aggarwal et al. Two birds with one network: Unifying failure event prediction and time-to-failure modeling
US11507880B2 (en) Automatic and continuous monitoring and remediation of API integrations
US20200166921A1 (en) System and method for proactive repair of suboptimal operation of a machine
US20160283310A1 (en) Anomaly classification, analytics and resolution based on annotated event logs
US20210097431A1 (en) Debugging and profiling of machine learning model training
Nigenda et al. Amazon sagemaker model monitor: A system for real-time insights into deployed machine learning models
US11449798B2 (en) Automated problem detection for machine learning models
US11755955B2 (en) Anomaly detection and tuning recommendation system
US11150631B2 (en) Predictive maintenance utilizing supervised sequence rule mining
CN114064196A (zh) 用于预测性保障的系统和方法
US20210158220A1 (en) Optimizing accuracy of machine learning algorithms for monitoring industrial machine operation
US11468365B2 (en) GPU code injection to summarize machine learning training data
WO2021067385A1 (fr) Débogage et profilage d'apprentissage de modèle d'apprentissage machine
US20230236923A1 (en) Machine learning assisted remediation of networked computing failure patterns
Harutyunyan et al. Challenges and Experiences in Designing Interpretable KPI-diagnostics for Cloud Applications
Khan Toward an Automated Real-Time Anomaly Detection Engine in Microservice Architectures

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: UNKNOWN

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

17P Request for examination filed

Effective date: 20210224

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

DAV Request for validation of the european patent (deleted)
DAX Request for extension of the european patent (deleted)
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: EXAMINATION IS IN PROGRESS

17Q First examination report despatched

Effective date: 20211202

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: SIEMENS MOBILITY AUSTRIA GMBH