[go: up one dir, main page]

WO2023035009A1 - Procédé et système pouvant être expliqués par des données pour une maintenance prédictive - Google Patents

Procédé et système pouvant être expliqués par des données pour une maintenance prédictive Download PDF

Info

Publication number
WO2023035009A1
WO2023035009A1 PCT/US2022/075974 US2022075974W WO2023035009A1 WO 2023035009 A1 WO2023035009 A1 WO 2023035009A1 US 2022075974 W US2022075974 W US 2022075974W WO 2023035009 A1 WO2023035009 A1 WO 2023035009A1
Authority
WO
WIPO (PCT)
Prior art keywords
controller
vehicle
neural network
event
data signals
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.)
Ceased
Application number
PCT/US2022/075974
Other languages
English (en)
Inventor
Eduardo Jose RAMIREZ LLANOS
Julien Ip
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.)
Continental Automotive Systems Inc
Original Assignee
Continental Automotive Systems Inc
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 Continental Automotive Systems Inc filed Critical Continental Automotive Systems Inc
Publication of WO2023035009A1 publication Critical patent/WO2023035009A1/fr
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/04—Inference or reasoning models
    • G06N5/045—Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence
    • 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
    • G05B23/0224—Process history based detection method, e.g. whereby history implies the availability of large amounts of data
    • G05B23/024—Quantitative history assessment, e.g. mathematical relationships between available data; Functions therefor; Principal component analysis [PCA]; Partial least square [PLS]; Statistical classifiers, e.g. Bayesian networks, linear regression or correlation analysis; Neural networks
    • 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
    • G05B23/0275—Fault isolation and identification, e.g. classify fault; estimate cause or root of failure
    • G05B23/0281—Quantitative, e.g. mathematical distance; Clustering; Neural networks; Statistical analysis
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning
    • G06N20/20—Ensemble 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
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/0464—Convolutional networks [CNN, ConvNet]
    • 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
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound

Definitions

  • the invention is directed to a method and system for predicting failures of a vehicle fleet, and particularly to a system and method for improving user friendliness associated with predictive maintenance.
  • Example embodiments of the present disclosure address shortcomings with existing machine learning by providing a predictive management system for one or more machines.
  • the system includes a controller having a plurality of input terminals coupled to receive a plurality of data signals from the one or more machines.
  • the controller is configured to predict a future event of the one or more machines from a plurality of future events based upon the received data signals, the plurality of future events including a plurality of failure events.
  • the controller is further configured to, upon the predicted future event being a first failure event of the plurality of failure events, send instructions to a user interface to display an identification of the first failure event, an explanation of the first failure event, and at least one data signal of the plurality of data signals during a defined time interval in which the at least one data signal is associated with the first failure event.
  • the controller includes a trained neural network and a machine learning algorithm executed by the trained neural network. The controller receives the plurality of data signals.
  • the neural network includes one or more output layers which identifies the predicted future event, a segmentation mask which determines the defined time interval associated with the predicted future event provided by the one or more output layers, and a determining block which, based upon the defined time interval, determines the explanation of the first failure event in a human readable form.
  • the neural network includes at least one of a convolutional neural network or a residual neural network having a plurality of layers and an average pooling layer.
  • the average pooling layer is downstream of the plurality of layers, relative to a direction of signal flow through the neural network.
  • the segmentation mask receives data associated with the average pooling layer.
  • the determining block includes a decision tree or a random forest which receives output of the segmentation mask and provides the human readable explanation based upon the segmentation mask output.
  • the one or more machines includes a vehicle from a fleet of one or more vehicles, and the plurality of data signals includes a plurality of sensor signals from sensors associated with at least one of an engine of the vehicle, an engine exhaust system of the vehicle, or at least one battery of the vehicle.
  • the plurality of data signals include sensor data signals
  • the controller includes inputs which receive the sensor data signals
  • the predictive management system further includes a user interface communicatively coupled to the controller, the user interface presenting the human readable explanation on a display.
  • controller and the display are remotely located from the vehicle.
  • a predictive management method for at least one machine includes receiving, at a controller, a plurality of data signals from the at least one machine.
  • the controller predicts a future event of the at least one machine from a plurality of future events based upon the received data signals, the plurality of future events including a plurality of failure events.
  • the controller determines a defined time interval in which at least one data signal of the plurality of data signals represents an anomaly associated with the first failure event, and determines a human readable explanation of the first failure event based in part upon the at least one data signal during the defined time interval, and sends instructions to a user interface to display an identification of the first failure event, the human readable explanation of the first failure event and the at least one data signal of the plurality of data signals during the defined time interval.
  • the controller includes a trained neural network executing a machine learning algorithm which receives the plurality of data signals.
  • the neural network includes one or more output layers which identifies the predicted future event.
  • the controller further includes a segmentation mask which determines the defined time interval associated with the predicted future event, and a determining block which, based upon the defined time interval, determines the human readable explanation associated with the first failure event.
  • the neural network may include at least one of a convolutional neural network or a residual neural network and may include a plurality of layers and an average pooling layer, the average pooling layer being downstream of the plurality of layers, relative to a direction of signal flow through the neural network, and the segmentation mask receives data associated with the average pooling layer.
  • the determining block may include a decision tree or a random forest which receives output of the segmentation mask and determines the human readable explanation based upon the segmentation mask output.
  • the at least one machine may include at least one vehicle
  • the predictive management method may include a predictive vehicle management method.
  • the plurality of data signals may include a plurality of sensor signals from sensors associated with an engine of the vehicle, with an engine exhaust system of the vehicle, with one or more tires of the vehicle, or with at least one battery of the vehicle.
  • controller and the user interface are remotely located from the vehicle. In another aspect, the controller and the user interface are located within the vehicle.
  • Fig. 1 illustrates example inputs and outputs of a conventional fleet maintenance system corresponding to predicted normal operation.
  • FIG. 2 illustrates example inputs and outputs of a conventional fleet maintenance system corresponding to an operation in which a failure event is predicted.
  • Fig. 3 illustrates example inputs and outputs of a fleet maintenance system corresponding to operation in which a failure event is predicted, according to an example embodiment.
  • Fig. 4 illustrates a schematic diagram of a fleet maintenance system controller according to an example embodiment capable of generating the outputs of Fig. 3.
  • Fig. 5 illustrates a schematic diagram of a fleet maintenance system according to an example embodiment in association with a fleet vehicle.
  • Example embodiments are directed to Al and machine learning provide utilizing methodologies to extract complex patterns from data and make predictions of future behaviors.
  • the methods and systems of the example embodiments provide a human readable and/or understandable explanation of the decision-making process concerning the future behavior. Provided explanations bring more transparency which translates in a more user-friendly process.
  • the example embodiments will be described as part of a predictive maintenance system for a fleet of one or more vehicles. It is understood, however, that the example embodiments are applicable in other deep learning or machine learning systems for predicting future machine behavior.
  • FIGs. 1-2 show an example of how an anomaly is presented to a fleet operator using a conventional predictive fleet maintenance system.
  • Fig. 1 shows inputs (depicted on the left side of the figure) and a corresponding output (appearing on the right side of the figure) for a fleet vehicle under normal operation
  • Fig. 2 shows an increased chance of a failure event (i.e., a “failure of actuator X”) predicted for the particular fleet vehicle.
  • the pertinent input signals are depicted over time for a particular fleet vehicle, and the output shows the probability of occurrence of any of a number of possible failures in a particular time interval T1 (represented in the shaded rectangle of the inputs).
  • the probability of occurrence of a particular failure event is calculated using a machine learning (ML) algorithm.
  • ML machine learning
  • Example embodiments address the above shortcoming of the conventional ML based fleet management system by providing additional information to the system user or other fleet operator. For instance, following the detection by the fleet management system of an anomaly during operation of a fleet vehicle, the anomaly is precisely located in a specific time interval T2 within the larger time interval T1 and an explanation of the anomaly is generated and presented to the fleet system operator in addition to the output in which a failure event (in this case, a failure of actuator X) is predicted.
  • Fig. 3 shows the inputs (variables) corresponding to the anomaly and the specific time interval T2 in which the anomaly occurs.
  • the output of the fleet management system includes the predicted failure event along with a calculated failure probability as part of a bar graph, and an explanation window 30 providing a brief explanation for the predicted failure event.
  • the extent of the explanation and the size of the window 30 may vary depending upon the amount of information believed to be needed to sufficiently inform the fleet system operator of the predicted failure of the fleet vehicle.
  • Fig. 4 is a schematic illustration of the architecture of at least part of a system controller 40 of a ML based predictive management system according to an example embodiment.
  • the particular application of the predictive management system is for managing a fleet of at least one vehicles.
  • the algorithm of the predictive management system controller 40 receives as inputs multi-variable time series signals.
  • the algorithm may also receive as inputs data labeled for training, validating and/or testing for supervised learning.
  • the algorithm performed by the system controller 40 provides the output shown in Fig.
  • the outputs may be provided to a display screen of the system that forms part of a user interface.
  • the predictive management system controller 40 is based on or otherwise includes a trained neural network 42, which may be a convolutional neural network (CNN) and/or a residual neural network (ResNet).
  • the system controller 40 is depicted in Fig. 4 as including a CNN but it is understood that a ResNet may alternatively or additionally be utilized.
  • the CNN illustrated is for one input variable for simplicity and it is understood the CNN may receive many input variables.
  • the CNN includes a plurality of convolutional layers 44 typically found in neural networks and/or CNNs.
  • An average pooling layer 45 is downstream of the convolutional layers 44.
  • the average pooling layer 45 averages the data of one or more of the convolutional layers 45 to obtain average pooling data.
  • the average pooling data is provided to a fully connected neural network and/or output layers 46, the last layer of which is a soft max layer which outputs the k output classes corresponding to the various predicted possible events of the fleet vehicle.
  • the various predicted events that are possible are used in forming a bar graph or the like for display, as shown in Fig. 3.
  • the various predicted possible events are normalized and the class (event) or classes having the largest magnitude form a basis of the predicted future behavior of the corresponding fleet vehicle, which may be a specific failure event or a no failure event.
  • the pooling layer 45 is a maximum (max) pooling layer in which
  • the fleet management system controller 40 includes a segmentation mask 48 which receives the average pooling data and determines the specific time interval T2 within the larger time interval T1 in which the variable pattern corresponding to the anomaly occurs.
  • the segmentation mask 48 is trained and is part of the controller 40.
  • the fleet management system controller 40 further includes an explanation determining block or module 50 which provides the explanation associated with the predicted event (failure or no failure) for display in the explanation window 30.
  • the explanation determining block 50 is a decision tree to split data according to certain parameters.
  • the explanation determining block 50 is a random forest.
  • the explanation determining block 50 is trained to identify possible explanations based upon the output of the segmentation mask 48, i.e., the values of the variable(s) in the specific time interval T2 which are associated with the anomaly resulting in the failure event prediction.
  • the decision tree of the explanation determining block 50 uses the segmentation mask for each variable, which facilitates its training.
  • the decision tree of the explanation determining block 50 may be built using known facts about each class and the segmentation mask 48. For example, if the output of the network 42, 46 points to an actuator for a diesel particulate filter (DPF) and the segmentation mask output shows lower than normal DPF or engine exhaust temperatures in the specific time window T2, then an explanation identifies in the explanation block 30 a low temperature being too low for effective diesel exhaust aftertreatment.
  • DPF diesel particulate filter
  • Another more elaborate way to provide explanations is by comparing the output statistics with normal operation statistics within a time window.
  • the explanation is generated by checking which values are in abnormal regions of their corresponding probability distribution.
  • Fig. 5 illustrates an implementation of the fleet management system according to an example embodiment in association with a fleet vehicle.
  • the vehicle fleet may include a plurality of fleet vehicles, only one fleet vehicle is shown for reasons of simplicity.
  • the fleet vehicle includes an engine (not shown), engine exhaust system (not shown) and a sensor system having a plurality of sensors which are operatively coupled to the engine and/or the engine exhaust system of the vehicle.
  • the sensors provide sensor data to a vehicle controller of the fleet vehicle.
  • the sensor data is stored in non-transitory memory of the vehicle controller.
  • An input/output (VO) block is communicatively coupled to the vehicle controller and includes circuitry for transmitting and receiving information wirelessly and/or over a physical connection. It is understood that the wired and wireless communication may utilize any known or future communication protocols and/or technologies.
  • the fleet management system controller 40 is separate and remotely located from each vehicle of the vehicle fleet. It is understood, however, that the fleet management system controller 40 may be located in a fleet vehicle and/or in each fleet vehicle.
  • the system controller 40 is communicatively coupled to an I/O block 51 having circuitry for transmitting data to and receiving data from the I/O block of each fleet vehicle, such as over the air interface or a wired, physical interface.
  • a user interface 52 is coupled to the system controller 40 and receives instructions to, among other things, display information depicted in Fig. 3.
  • the user interface 52 is also configured to receive system operator input.
  • the user interface may include a display, such as a touch screen display, a keyboard/keypad, mousejoystick, electronic pen, etc.
  • system controller 40 may further include additional data processing hardware (e.g., data processing hardware and/or a central processing unit having one or more computing processors) in communication with non-transitory memory or memory hardware (e.g., a hard disk, flash memory, random-access memory) capable of storing instructions executable on the computing processor(s)).
  • additional data processing hardware e.g., data processing hardware and/or a central processing unit having one or more computing processors
  • non-transitory memory or memory hardware e.g., a hard disk, flash memory, random-access memory
  • the memory hardware may store additional software program code having instructions which, when executed by the data processing hardware, may perform one or more operations or functions that supports or is otherwise related to the predictive maintenance operations described herein.
  • the fleet management system controller 40 has been described for use in predicting failure events of an engine and/or engine exhaust system of one or more fleet vehicles. It is understood that the system controller 40 also may be used in predicting failure events for virtually any type of machine.
  • the system controller 40 may be used in predicting failure events for an electric vehicle or a fleet of electric vehicles.
  • the data signals received from the electric vehicle may include sensor data such as battery temperature and battery voltage, and one failure event may be battery overheating and the batteries fail to sufficiently charge.
  • the system controller 40 may be used in predicting failure events for tires of a vehicle based upon sensors associated with the vehicle tires. The vehicle tire sensors may sense tire rotation, tire pressure, etc.
  • system controller 40 may, based on the tire sensor data, predict the onset of imminent tire failure due to, for example, thread separation at higher speeds.
  • the system controller 40 may receive sensor output data associated with any and/or all components, assemblies, subassemblies, etc. of a vehicle or fleet thereof for use in predicting failures of or otherwise managing the vehicle or vehicle fleet.
  • the fleet management system controller 40 includes a machine learning algorithm having a trained model that is built on training data.
  • the trained model is able to make predictions or decisions without explicit programming code to do so.
  • the neural network 42 pertains to computational approaches used in computer science, among other disciplines, and is based on a large collection of neural unites, loosely imitating the way a biological brain solves problems with large clusters of biological neurons connected by axons.
  • the neural network 42 is self-learning and trained, rather than programed, and excels in areas where the solution feature detection is difficult to express in a traditional computer program. In other words, the neural network 42 is a set of algorithms that is designed to recognize patterns.
  • the recognized patterns are numerical, vectors, into which all-real-world data, such as images, text, sound, or time series is translated.
  • the neural network 42 includes multiple layers 44 of nonlinear processing units in communication with non-transitory memory.
  • the non-transitory memory stores instructions that when executed on the nonlinear processing units cause the neural network 42 to provide an output.
  • Each nonlinear processing unit is configured to transform an input or signal using parameters that are learned through training. A series of transformations from input to outputs occur at the multiple layers 44 of the nonlinear processing units.
  • the particular use of the neural network 42 is described above for predicting a future event associated with the engine or engine exhaust system of one or more vehicles.
  • the neural network 42 forms part of the predictive management system controller 40 for predicting a future event, such as a failure event, based on sensor and/or other data from, for example, the engine and/or engine exhaust system of each fleet vehicle of a plurality of vehicles in a vehicle fleet.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Data Mining & Analysis (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Computational Linguistics (AREA)
  • Biophysics (AREA)
  • Health & Medical Sciences (AREA)
  • Medical Informatics (AREA)
  • Molecular Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Biomedical Technology (AREA)
  • Automation & Control Theory (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Mathematical Analysis (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Pure & Applied Mathematics (AREA)
  • Mathematical Optimization (AREA)
  • Algebra (AREA)
  • Probability & Statistics with Applications (AREA)
  • Combined Controls Of Internal Combustion Engines (AREA)

Abstract

Un système et un procédé de gestion prédictive pour une machine sont divulgués. Un dispositif de commande reçoit un signal de données de machine et est configuré pour prédire un événement futur de la machine à partir d'une pluralité d'événements futurs sur la base des signaux de données reçus. Le dispositif de commande, lorsque l'événement futur prédit est un premier événement de défaillance, envoie également des instructions à une interface utilisateur pour afficher une identification et une explication du premier événement de défaillance, et au moins un signal de données pendant un intervalle de temps défini dans lequel le signal de données est associé au premier événement de défaillance. Le dispositif de commande comprend un réseau de neurones artificiels entraîné qui reçoit le signal de données et comprend une ou plusieurs couches de sortie identifiant l'événement futur prédit, un masque de segmentation qui détermine l'intervalle de temps défini, et un bloc de détermination qui, sur la base de l'intervalle de temps défini, détermine l'explication associée au premier événement de défaillance sous une forme lisible par l'homme.
PCT/US2022/075974 2021-09-03 2022-09-06 Procédé et système pouvant être expliqués par des données pour une maintenance prédictive Ceased WO2023035009A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US202163260893P 2021-09-03 2021-09-03
US63/260,893 2021-09-03

Publications (1)

Publication Number Publication Date
WO2023035009A1 true WO2023035009A1 (fr) 2023-03-09

Family

ID=83508614

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2022/075974 Ceased WO2023035009A1 (fr) 2021-09-03 2022-09-06 Procédé et système pouvant être expliqués par des données pour une maintenance prédictive

Country Status (1)

Country Link
WO (1) WO2023035009A1 (fr)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019185657A1 (fr) * 2018-03-27 2019-10-03 We Predict Limited Procédé de diagnostic de véhicule prédictif
US20190325328A1 (en) * 2018-04-19 2019-10-24 Ptc Inc. Detection and use of anomalies in an industrial environment
EP3637083A1 (fr) * 2018-10-11 2020-04-15 Hyundai Motor Company Procédé de diagnostic de pannes pour composants de transmission
WO2020110718A1 (fr) * 2018-11-30 2020-06-04 いすゞ自動車株式会社 Dispositif et procédé de création de modèle, et programme
EP3667447A1 (fr) * 2018-12-12 2020-06-17 Hyundai Motor Company Procédé et dispositif de diagnostic d'une source de bruit problématique sur la base d'informations de données de grande taille

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2019185657A1 (fr) * 2018-03-27 2019-10-03 We Predict Limited Procédé de diagnostic de véhicule prédictif
US20190325328A1 (en) * 2018-04-19 2019-10-24 Ptc Inc. Detection and use of anomalies in an industrial environment
EP3637083A1 (fr) * 2018-10-11 2020-04-15 Hyundai Motor Company Procédé de diagnostic de pannes pour composants de transmission
WO2020110718A1 (fr) * 2018-11-30 2020-06-04 いすゞ自動車株式会社 Dispositif et procédé de création de modèle, et programme
EP3667447A1 (fr) * 2018-12-12 2020-06-17 Hyundai Motor Company Procédé et dispositif de diagnostic d'une source de bruit problématique sur la base d'informations de données de grande taille

Similar Documents

Publication Publication Date Title
US8332337B2 (en) Condition-based monitoring system for machinery and associated methods
CN107402921B (zh) 识别用户行为的事件时序数据处理方法、装置及系统
US8849732B2 (en) Adaptive remote maintenance of rolling stocks
KR102215107B1 (ko) 주행 데이터 기반 차량 상태 예측 시스템 및 방법
CN119087962B (zh) 一种基于人工智能的车辆故障诊断和维护评估方法
US12272121B2 (en) Image processing device, system, and method
Thandlam et al. Next-gen vehicle safety: The futuristic approach to auto-stop in modern vehicles
US20220067535A1 (en) Anomaly detection in cyber-physical systems
CN118041762A (zh) 工业控制系统的通信网络故障预测方法及装置
CN114244596A (zh) 基于htm的车载can网络异常检测方法和系统
CN115496384A (zh) 工业设备的监控管理方法、装置和计算机设备
CN114662554A (zh) 用于诊断设备的系统和方法以及诊断设备的组件的方法
CN119990250A (zh) 车辆故障预测模型的训练方法、装置及车辆故障预测方法
CN120481775A (zh) 智能储能电源的远程故障诊断方法及系统
WO2022258292A1 (fr) Procédé mis en œuvre par ordinateur pour la commande d'une classification effectuée par un réseau neuronal en graphes, unité de commande et produit-programme informatique
CN118194156A (zh) 知识和数据融合驱动迭代的核电站水泵故障诊断方法
WO2023035009A1 (fr) Procédé et système pouvant être expliqués par des données pour une maintenance prédictive
CN119402337B (zh) 一种告警数据的处理方法、装置、设备及介质
CN120046778A (zh) 用于钢铁生产工艺参数优化的智能化综合系统及方法
Jang et al. Anomaly detection of 2.4 l diesel engine using one-class SVM with variational autoencoder
CN112861930A (zh) 基于机器学习的车辆故障分析及预测方法和装置
Vachtsevanos et al. Prognosis: Challenges, precepts, myths and applications
CN119051882B (zh) 一种基于深度学习的智能网联汽车车载网络异常检测方法
Tod et al. Augmenting train maintenance technicians with automated incident diagnostic suggestions
Mumcuoglu et al. Air pressure system failures detection using LSTM-autoencoder

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22783232

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22783232

Country of ref document: EP

Kind code of ref document: A1