Systems and methods for remaining useful life (RUL) detection and control based on the detection

Abstract

Systems and methods for detecting a health state and / or RUL of a vehicle component can include one or more processors that obtain a plurality of data sets of one or more vehicles such that each data set is recorded from a corresponding vehicle at a corresponding RUL of a vehicle component and includes parameter values of a parameter set of the vehicle. The one or more processors assign a corresponding health state of the vehicle component to each data set based on the corresponding RUL and select a subset of parameters of the parameter set based on an influence of the parameters on the health state and / or RUL of the vehicle component. The one or more processors train a machine learning model for detecting the health state and / or RUL of the vehicle component using values of the subset of parameters and provide the machine learning model for deployment.

Description

[0001] Cross-references to related applications This PCT application claims the benefit and priority of Indian Patent Application No. 202441000823, filed on 4 January 2024, entitled SYSTEMS AND METHODS FORREMAINING USEFUL LIFE (RUL) DETECTION AND CONTROLS BASED THEREON, which is incorporated herein by reference in its entirety and for all purposes. Technical Field

[0002] This disclosure relates to monitoring the health status of systems and / or components of vehicles and / or other equipment. More specifically, this disclosure relates to systems and methods for detecting the remaining useful life (RUL) of components or systems of a vehicle or equipment using at least one machine learning model. The system and method may include: developing a virtual RUL sensor using one or more machine learning models; and using the virtual RUL sensor to detect the RUL or health status of one or more systems or components of the vehicle or equipment. Background Technology

[0003] In recent years, emission regulations for internal combustion engines have become increasingly stringent. Concerns about environmental issues have driven many regions worldwide to implement stricter emission requirements for internal combustion engines. Government agencies rigorously monitor engine emissions and establish emission standards that engines must comply with. At least some emission standards require proper and timely maintenance of the vehicle's or equipment's aftertreatment system, engine system, transmission system, and / or other systems.

[0004] When systems, components, and / or parts of a vehicle or equipment age, they may fail to operate in accordance with existing emission standards and / or other standards or regulations. Specifically, the performance of systems, components, and / or parts of a vehicle or equipment typically degrades gradually over time. As performance degradation intensifies over time, for example, gradually and / or abruptly, the system, component, and / or part may reach a point of failure (or failure state) where its performance no longer meets the given standards, regulations, and / or specifications set by government agencies and / or the vehicle (or equipment) or its system manufacturer. For example, the performance of a treatment system or its components may degrade to the point where the vehicle (or equipment) emissions do not meet one or more pints set by the equipment manufacturer. Summary of the Invention

[0005] One embodiment relates to a system for training a machine learning model for detecting the health status of a vehicle component. The system may include a memory and one or more processors, the memory storing executable instructions. When executed by the one or more processors, these executable instructions cause the one or more processors to obtain multiple datasets of one or more vehicles. Each dataset is recorded from a corresponding vehicle among the one or more vehicles at a corresponding remaining useful life (RUL) of the vehicle component and includes parameter values ​​of a predefined set of parameters for that vehicle. The one or more processors may: assign a corresponding health status of the vehicle component to each of the multiple datasets based on the corresponding RUL of the vehicle component; select a subset of parameters from the predefined parameter set based on changes in the predefined parameter set in response to changes in the health status of the vehicle component; train a machine learning model for detecting the health status of the vehicle component using the parameter values ​​of the subset of parameters from the multiple datasets; and provide the machine learning model for detecting the health status of the vehicle component.

[0006] In some implementations, the vehicle component is a diesel oxidation catalyst (DOC). This subset of parameters may include at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in the aftertreatment, Maximum Combined Soot Load (CSLE) parameter in the aftertreatment, Maximum MBSLE parameter, Average Pressure Difference Soot Load Estimate (DPSLE) offset adjustment parameter, or Maximum Urea Pump Pressure (related to DEF pressure) parameter.

[0007] In some implementations, the vehicle component is a selective catalytic reduction (SCR) system. This subset of parameters may include at least one of the following: maximum CSLE parameter, maximum MBSLE parameter, parameter representing the average urea pump command when pressure is in closed loop but without metered injection, maximum diesel particulate filter (DPF) differential pressure parameter, or parameter representing the average compressor inlet pressure when the equipment (or vehicle) is operating, which is the fifth most predictive parameter.

[0008] In some implementations, the one or more processors are configured to select a plurality of parameters having the largest range of variation in response to a change in the health state of the vehicle component. The one or more processors are configured to use an extreme gradient boosting (XGBoost) algorithm to select the plurality of parameters having the largest range of variation in response to a change in the health state of the vehicle component.

[0009] In some implementations, the machine learning model includes at least one of the following: a random forest model; a statistical learning model; or a neural network.

[0010] In some implementations, the health state of the vehicle component includes an unhealthy state when the RUL of the vehicle component is less than a predefined threshold and a healthy state when the RUL of the vehicle component is greater than the predefined threshold. The predefined threshold can be defined as the additional distance that the vehicle associated with the vehicle component must travel before the vehicle component's performance reaches a failure state.

[0011] Another embodiment relates to a method for training a machine learning model for detecting the health status of a vehicle component, the method comprising: obtaining, by a computer system, a plurality of datasets relating to one or more vehicles, each dataset being a record from a corresponding vehicle among the one or more vehicles at a corresponding remaining useful life (RUL) of the vehicle component, and including parameter values ​​of a predefined set of parameters for the vehicle; assigning a corresponding health status of the vehicle component to each of the plurality of datasets based on the corresponding RUL of the vehicle component; selecting a subset of parameters of the predefined set using the plurality of datasets based on changes in the predefined set in response to changes in the health status of the vehicle component; training a machine learning model for detecting the health status of the vehicle component using the parameter values ​​of the subset of parameters in the plurality of datasets; and providing the machine learning model for detecting the health status of the vehicle component by the computer system.

[0012] In some implementations, the vehicle component is a diesel oxidation catalyst (DOC). This subset of parameters may include at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in the aftertreatment, Maximum Combined Soot Load (CSLE) parameter in the aftertreatment, Maximum MBSLE parameter, Average Pressure Difference Soot Load Estimate (DPSLE) offset adjustment parameter, or Maximum Urea Pump Pressure (related to DEF pressure) parameter.

[0013] In some implementations, the vehicle component is a selective catalytic reduction (SCR) system. This subset of parameters may include at least one of the following: maximum CSLE parameter, maximum MBSLE parameter, parameter representing the average urea pump command when pressure is in closed loop but without metered injection, maximum diesel particulate filter (DPF) differential pressure parameter, or parameter representing the average compressor inlet pressure when the equipment (or vehicle) is operating, which is the fifth most predictive parameter.

[0014] In some implementations, the method includes selecting a plurality of parameters having the largest range of variation in response to a change in the health state of the vehicle component. The method further includes using an extreme gradient boosting (XGBoost) algorithm to select the plurality of parameters having the largest range of variation in response to a change in the health state of the vehicle component.

[0015] In some implementations, the machine learning model includes at least one of the following: a random forest model; a statistical learning model; or a neural network.

[0016] In some implementations, a vehicle component is defined as being in an unhealthy state if its RUL (Range Limit Indicator) is less than a predefined threshold. The predefined threshold is defined as the additional distance the vehicle corresponding to that component must travel before the component's performance reaches the failure criterion.

[0017] Another embodiment relates to a system for detecting the health status of vehicle components. The system includes a memory and one or more processors. The memory stores executable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations including: receiving parameter values ​​of a plurality of parameters of the vehicle; using a machine learning model and the parameter values ​​to determine the health status of a vehicle component, the health status indicating whether the remaining useful life (RUL) of the vehicle component is less than a predefined threshold; and providing an indication of the health status for display on a user interface.

[0018] In some implementations, the vehicle component is a diesel oxidation catalyst (DOC). This subset of parameters includes at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in the aftertreatment, Maximum Combined Soot Load (CSLE) parameter in the aftertreatment, Maximum MBSLE parameter, Average Pressure Difference Soot Load Estimation (DPSLE) offset adjustment parameter, or Maximum Urea Pump Pressure (related to DEF pressure) parameter.

[0019] In some implementations, the vehicle component is a selective catalytic reduction (SCR) system. This subset of parameters includes at least one of the following: maximum CSLE parameter, maximum MBSLE parameter, parameter representing the average urea pump command when pressure is in closed loop but without metered injection, maximum diesel particulate filter (DPF) differential pressure parameter, or parameter representing the average compressor inlet pressure when the equipment (or vehicle) is operating, which is the fifth most predictive parameter.

[0020] In some implementations, the machine learning model includes at least one of the following: a random forest model; a statistical learning model; or a neural network. A healthy state is considered unhealthy if the RUL of the vehicle component is less than a predefined threshold.

[0021] Another embodiment relates to a method for detecting the health status of a vehicle component. The method includes: receiving parameter values ​​of a plurality of parameters of the vehicle by one or more processors; determining, by the one or more processors, the health status of a vehicle component using a machine learning model and the parameter values, the health status indicating whether the remaining useful life (RUL) of the vehicle component is less than a predefined threshold; and providing an indication of the health status by the one or more processors for display on a user interface.

[0022] In some implementations, the vehicle component is a diesel oxidation catalyst (DOC). This subset of parameters includes at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in the aftertreatment, Maximum Combined Soot Load (CSLE) parameter in the aftertreatment, Maximum MBSLE parameter, Average Pressure Difference Soot Load Estimation (DPSLE) offset adjustment parameter, or Maximum Urea Pump Pressure (related to DEF pressure) parameter.

[0023] In some implementations, the vehicle component is a selective catalytic reduction (SCR) system. This subset of parameters includes at least one of the following: maximum CSLE parameter, maximum MBSLE parameter, parameter representing the average urea pump command when pressure is in closed loop but without metered injection, maximum diesel particulate filter (DPF) differential pressure parameter, or parameter representing the average compressor inlet pressure when the equipment (or vehicle) is operating, which is the fifth most predictive parameter.

[0024] In some implementations, the machine learning model includes at least one of the following: a random forest model; a statistical learning model; or a neural network. A healthy state is considered unhealthy if the RUL of the vehicle component is less than a predefined threshold.

[0025] The present invention is merely illustrative and is not intended to be limiting in any way. Other aspects, inventive features, and advantages of the apparatus or process described herein will become apparent from the detailed description set forth herein in conjunction with the accompanying drawings, wherein like reference numerals refer to like elements. Numerous specific details are provided to provide a thorough understanding of embodiments of the subject matter of this disclosure. The described features of the subject matter of this disclosure may be combined in any suitable manner in one or more embodiments and / or implementations. In this regard, one or more features of one aspect of the invention may be combined with one or more features of different aspects of the invention. Furthermore, additional features may be recognized in some embodiments and / or implementations and are not present in all embodiments or implementations. Attached Figure Description

[0026] Figure 1 This is a schematic diagram of a computing and networking environment for detecting the RUL and / or health status of components and / or systems in a vehicle and / or equipment, according to an example embodiment.

[0027] Figure 2 According to the example embodiment Figure 1 A schematic diagram of the controller for the RUL and / or health status monitoring system.

[0028] Figure 3 This is a flowchart of a method for generating or training a virtual RUL or health status detector according to an example embodiment.

[0029] Figure 4 This is a flowchart depicting the acquisition of data from a vehicle or equipment according to an example embodiment.

[0030] Figure 5 This is a diagram illustrating the determination of the RUL and health status of vehicle components from recorded data according to an example embodiment.

[0031] Figure 6 It is a graph depicting the relationship between the SHAP values ​​of a plurality of selected vehicle parameters according to an example embodiment and the health status of a vehicle component or system (shown as a diesel oxidation catalyst (DOC)).

[0032] Figure 7 It is a graph depicting the relationship between the SHAP values ​​of a plurality of selected vehicle parameters according to an example embodiment and the health status of a vehicle component or system (shown as a selective catalytic reduction (SCR) system).

[0033] Figure 8 It is a graph depicting the relationship between the SHAP values ​​of a plurality of selected vehicle parameters according to an example embodiment and the health status of a vehicle component or system (shown as a particulate filter, and specifically a diesel particulate filter (DFP)).

[0034] Figure 9 This is based on the use of the example embodiment. Figure 3 A flowchart of a method for detecting the health status and / or RUL of vehicle and / or equipment components using virtual health status and / or RUL detectors generated in the system. Detailed Implementation

[0035] The following describes in more detail various concepts, apparatuses, and systems related to generating one or more virtual remaining useful life (RUL) detectors or sensors, and methods, apparatuses, and systems for detecting the RUL of one or more systems and / or components of a vehicle or equipment using virtual RUL detectors or sensors. Before turning to the accompanying drawings, which illustrate certain exemplary embodiments in detail, it should be understood that this disclosure is not limited to the details or methods set forth in the specification or illustrated in the drawings. It should also be understood that the terminology used herein is for descriptive purposes only and should not be considered limiting.

[0036] For example, emission standards or regulations established by government agencies set specific requirements for gases emitted by vehicles and / or other equipment. Vehicles and / or other equipment should meet these requirements during operation. However, the performance of components of vehicles and / or other equipment (e.g., components corresponding to aftertreatment systems) typically degrades over time and may fail or reach a failure state at some point. As used herein, a component failure, failure event, or failure state refers to a state in which the component's performance does not conform to predefined expected operating characteristics. In one embodiment, the predefined expected operating characteristics may correspond to at least one requirement or regulation. In another embodiment, the predefined expected operating characteristics may correspond to a value different from at least one requirement or regulation (e.g., a value less than a predefined maximum value to provide a buffer to help achieve the expected operation). Strengthening compliance with existing operating standards and / or existing requirements or regulations requires automatic or near-automatic monitoring of the health status of relevant components and / or systems of vehicles or equipment, and detection of the corresponding RUL (Rating of Limitations, Limitations, and Limitations).

[0037] Regular monitoring of the health status of components or systems over time, or the Remaining Duration and Limitation (RUL), allows for the early prediction of failure events and the taking of appropriate action before a component or system reaches a failure state. As used herein, the RUL of a component or system refers to the remaining runtime before a failure event occurs (which can be expressed as a predefined amount of time and / or predefined additional distance value that the corresponding vehicle would travel when the system, computer-readable medium, and method are applied to a vehicle). The remaining runtime can be the remaining operating time before a failure event occurs (e.g., during which the corresponding equipment or vehicle is turned on or is operating). Therefore, the smaller the RUL value, the closer the component and / or system is to likely experiencing a failure event. When a failure event of a component or system is determined to be "approaching," the detection system can send an alert to the vehicle or equipment operator, owner, and / or maintenance facility to request maintenance or replacement of the monitored component or system.

[0038] The systems, computer-readable media, and methods described herein enable virtual detectors to detect the relative health status (RUL) and / or health status of components or systems in a vehicle or equipment using at least one machine learning model. The health status of a component or system can be defined based on a corresponding RUL. For example, a healthy status can be defined as "unhealthy" when the RUL of a component or system is below a predefined threshold, and a healthy status can be defined as "healthy" when the RUL is above a predefined threshold. The systems, computer-readable media, and methods described herein can train at least one machine learning model to estimate the RUL of a component or system and / or predict its health status. The systems and methods described herein can train the machine learning model using a subset of parameters selected from a large set of available parameters for one or more vehicles or equipment. The selection can be based on changes in the parameters in response to changes in the health status and / or RUL of the component or system.

[0039] Once the machine learning model is trained, the systems and methods described herein can be deployed to monitor the near-ultimate health status (RUL) or health condition of components in one or more vehicles. The systems and methods described herein can acquire data corresponding to a selected subset of parameters from the vehicle or other equipment and feed that data as input to the trained machine learning model. In response, the machine learning model can provide the health condition and / or RUL of a component or system as output. The systems and methods described herein can provide or transmit the health condition and / or RUL of a component or system to dashboards, graphical user interfaces, and / or computing devices to trigger proactive maintenance actions before the component or system reaches a failure state.

[0040] While some of the example embodiments described below relate to RUL and / or health status detection for selective catalytic reduction (SCR) systems, particulate filters (such as diesel particulate filters (DPF)), and diesel oxidation catalysts (DOC), it should be noted that the methods and systems described herein for training machine learning models and / or detecting RUL or health status are applicable to other components of aftertreatment systems, or more generally to other components and / or systems of vehicles or equipment (e.g., generators). For example, similar methods for training or generating machine learning models or virtual sensors and / or using machine learning models or virtual sensors to detect component RUL and / or health status can be used for various types of components and / or systems, such as diesel exhaust fluid (DEF) meters, mufflers, power steering systems, catalytic converters, transmission systems, batteries, alternators, radiators, clutches, and brakes, etc.

[0041] From a technical and benefit perspective, the systems and methods described herein enable reliable and early detection / prediction of various components, parts, or systems in vehicles and / or equipment. Failures can be detected or predicted before they occur, allowing for early and proactive maintenance or repair. Similarly, machine learning-based methods allow for relatively high accuracy in fault detection / prediction and can be customized for various types of components, parts, or systems in vehicles and / or equipment. Furthermore, the systems and methods described herein provide computationally efficient fault predictors. Specifically, the machine learning models are trained or generated based on a selected set of parameters that are relatively more correlated with or more sensitive to potential future failures and are more effective in predicting future failure events. Choosing a relatively small number of parameters (rather than using the entire available set) reduces the computational complexity of the machine learning model and training process without compromising the effectiveness or reliability of the prediction or detection. Finally, early detection or prediction of failure events allows for early maintenance (e.g., before failure events occur), leading to increased uptime and extended service life of vehicles and / or equipment. These features and benefits, as well as others, are described more fully below.

[0042] Now for reference Figure 1 A schematic diagram of a computing and networking environment 100 for detecting the RUL and / or health status of components or systems in vehicles and / or equipment is shown, according to various example embodiments. In short, the computing and networking environment 100 may include a health status detection system 102, one or more vehicles (or equipment) 104, and one or more computer systems 106. The health status detection system 102 (also referred to herein as an RUL detection system or RUL / health status detection system, health status detection computing system, and / or provider computing system), one or more vehicles (or equipment) 104, and one or more computer systems 106 may be communicatively connected via a communication network 108. The health status detection system 102 may include data acquisition circuitry 110, a virtual detector (or sensor) generator 112, and one or more virtual RUL / health status detectors 114a to 104n (referred to herein individually or in combination as RUL / health status detector 114, health status detector 114, or RUL detector 114). The vehicle (or equipment) 104 may include a control system 116 and a telematics unit 118, as well as other components and / or systems. Computer system 106 may include a graphical user interface (or dashboard) for displaying information, such as data or notifications received from health status monitoring system 102.

[0043] Vehicle 104 can be a highway vehicle or an off-road vehicle, including but not limited to long-haul trucks, medium-duty trucks (e.g., pickup trucks), passenger cars (e.g., sedans, hatchbacks, coupes, etc.), buses, vans, garbage trucks, fire trucks, concrete mixer trucks, delivery trucks, locomotives, ships, aircraft, and other types of vehicles. In some embodiments, Figure 1 The vehicle 104 shown may be fixed or substantially fixed equipment, such as a generator, generator set, crane, oil drilling equipment, and / or other types of equipment. Generally speaking, the embodiments disclosed in this disclosure can be applied to vehicles and / or equipment.

[0044] Vehicle 104 may include: at least one control system or controller 116 configured or constructed to monitor and / or manage the performance of one or more systems or subsystems of the vehicle; and a telematics unit 118 for communicating with health status detection system 102 and / or other remote systems or devices.

[0045] Control system 116 may include any type of control system included in a vehicle, including but not limited to engine control modules or units (ECM or ECU), powertrain control modules, transmission control modules, after-treatment system control modules, and / or combinations thereof. Control system 116 may include processing circuitry having one or more processors and one or more memories. The processor may be implemented as a single-chip or multi-chip processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. The processor may be a microprocessor. The processor may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration. Memory 206 (e.g., memory cells and / or storage devices) may include one or more devices (e.g., RAM, ROM, flash memory, hard disk storage devices) for storing data and / or computer code used to perform or facilitate the various processes, layers, and modules described herein. Memory 206 may be communicatively connected to processor 204 to provide processor 204 with computer code or instructions for performing at least some of the procedures described herein. Furthermore, memory 206 may be or include tangible, non-transient volatile memory or non-volatile memory. Therefore, memory 206 may include database components, object code components, script components, or any other type of information structure to support the various activities and information structures described herein.

[0046] The control system 116 can provide commands and / or instructions (e.g., timing and amount of fuel injection, exhaust gas recirculation percentage, turbo compressor speed, etc.) to one or more actuators within the vehicle 104. Furthermore, the control system 116 is configured to receive, acquire, read, record, track, and / or interpret data from one or more components and / or systems within the vehicle 104. Therefore, the control system 116 can be communicatively coupled to one or more sensors included within the vehicle 104. For example, the control system 116 can receive engine speed data and engine torque data from each of an engine speed sensor and an engine torque sensor, respectively. The control system 116 can receive measurements from other sensors, such as one or more NOx sensors, diesel exhaust fluid (DEF) metering injection sensors, and / or exhaust system thermometers, and other vehicle sensors. For tracking, partitioning, and analysis, each piece of data can be associated with a data identifier (DID) (e.g., code, value, etc.). The control system 116 can also receive data from on-board diagnostic systems (e.g., OBD II, OBD I, EOBD, JOBD, etc.). Therefore, the control system 116 can receive diagnostic fault codes (DTCs) based on one or more operating characteristics of components in the vehicle 104. DTCs may include fault codes, parameter IDs, etc.

[0047] The telematics unit 118 is communicatively coupled to the control system 116 and the telematics platform in the vehicle 104, which is associated with or communicatively coupled to the health status monitoring system 102, for example. The telematics unit 118 can be configured as any type of telematics control unit. Therefore, the telematics unit 118 may include, but is not limited to: a positioning system (e.g., Global Positioning System) for tracking the vehicle's location (e.g., latitude and longitude data, altitude data, etc.), one or more memory devices for storing data, one or more electronic processing units for processing data, and a communication interface for facilitating data exchange between the telematics unit 118 and the telematics platform or the health status monitoring system 102. In this regard, the communication interface can be configured as any type of mobile communication interface or protocol, including but not limited to Wi-Fi, WiMAX, Internet, radio, Bluetooth, ZigBee, satellite, cellular, GSM, GPRS, LTE, etc. The telematics unit 118 may also include a communication interface for communicating with the control system 116 of the vehicle 104. The communication interface used to communicate with the control system 116 may include any type and number of wired and wireless protocols (e.g., any standard under IEEE 802, etc.). In some implementations, communication between the telematics unit 118 and the control system 116 may be conducted via the Unified Diagnostic Services (UDS) protocol.

[0048] In some embodiments, the telematics unit may be excluded or omitted, and the controller 116 may include or be coupled to a communication interface configured to enable communication with a remote computing system.

[0049] Components or systems of vehicle 104 can communicate with each other or with remote components using any type and any number of wired or wireless connections. For example, wired connections may include serial cables, fiber optic cables, CAT5 cables, or any other form of wired connection. Wireless connections may include the Internet, Wi-Fi, cellular, radio, Bluetooth, ZigBee, etc. In one embodiment, a Controller Area Network (CAN) bus provides the exchange of signals, information, and / or data. The CAN bus includes any number of wired and / or wireless connections. In some implementations, control system 116 communicates with other components of vehicle 104 via the CAN bus.

[0050] Computer system 106 may include any combination of computing devices, such as computer servers, desktops, laptops, tablets, and / or handheld devices. Computer system 106 may be associated with the owner of vehicle 104 and / or a third party monitoring the health status of one or more components of vehicle 104. For example, computer system 106 may be associated with the vehicle operator, original equipment manufacturer (OEM), entity managing a fleet of vehicles 104, and / or vehicle maintenance facility, etc. Computer system 106 may be communicatively coupled to health status monitoring system 102 and / or vehicle 104 via communication network 108. Communication network 108 may include the Internet, one or more wireless communication networks, one or more Wi-Fi networks, one or more local area networks (LANs), one or more wide area networks (WANs), a fixed-line telephone network, or a combination thereof.

[0051] Health status detection system 102 may include one or more computing devices, such as computer servers, desktops, laptops, tablets, and / or handheld devices. In some implementations, health status detection system 102 or portions thereof may reside in the cloud. According to such implementations, functionality or operation associated with health status detection system 102 may be implemented or performed via one or more hardware servers and / or one or more virtual servers. Health status detection system 102 includes data acquisition circuitry 110, a virtual RUL detector generator 112 (also referred to herein as health status detector generator 112 or RUL detector generator 112), and one or more health status detectors 114. Each of the components or circuits 110, 112, and 114 may be implemented as hardware, firmware, software, or a combination thereof. For example, each of the components or circuits 110, 112, and 114 may be implemented wholly or partially as executable instructions executed by one or more processors. Health status detection system 102 may be owned, operated, and / or managed by, or associated with, a provider organization (e.g., a service, equipment, etc.). For example, a provider can be an original equipment manufacturer (OEM). A provider can also offer services for tracking and maintaining the health of one or more components and / or systems. In this case, the provider could be an engine manufacturer that also provides various services.

[0052] Data acquisition circuitry 110 may be configured or constructed to acquire training data, validation data, and detection data. Data acquisition circuitry 110 may include a communication interface for communicating, for example, with vehicle 104, computer system 106, and / or other remote systems or devices via communication network 108. Data acquisition circuitry 110 may include one or more circuits for generating data requests and / or tracking the timing of data requests. Data acquisition circuitry 110 may be communicatively coupled to health status / RUL detector generator 112 and health status detector 114. Data acquisition circuitry 110 may receive training data (e.g., including validation (or test) data) from vehicle 104, computer system 106, or some other data source and provide the training data to health status detector generator 112. Data acquisition circuitry 110 may receive or acquire sensor data from vehicle 104 and provide the received sensor data to health status detector 114. Data acquisition circuitry 110 may include a memory or database for storing training data, validation (or test) data, and / or detection data.

[0053] Virtual health status / RUL detector generator 112 can be configured or constructed to generate one or more virtual health status / RUL detectors 114 for detecting the RUL or health status of one or more specific components and / or systems of one or more vehicles in vehicle 104. Specifically, virtual health status / RUL detector generator 112 may receive training data from data acquisition circuitry 110 or a database and use the training data to train one or more machine learning models. The training data may include parameter values ​​(or parameter measurements, if from one or more physical sensors) of a parameter set of vehicle 104 or its subsystems. Virtual health status / RUL detector generator 112 may receive validation (or test) data from data acquisition circuitry 110 or its database and use the validation (or test) data to validate or test the trained machine learning model. The validation data may be separate from and different from the training data. In some implementations, virtual health status / RUL detector generator 112 may use a first portion of the training data to train the machine learning model and a second portion to validate or test the trained model. The machine learning model is trained and constructed or configured to detect the health status or RUL of the corresponding vehicle components of vehicle 104 based on the parameter values ​​of vehicle 104.

[0054] The virtual health status / RUL detector generator 112 can preprocess training data before it is used to train a machine learning model. Preprocessing may include editing data, labeling datasets and / or filtering data, or selecting a subset of vehicle parameters whose values ​​will be used to train the machine learning model. The virtual health status / RUL detector generator 112 can apply the same preprocessing to validation (or test) data. The following section combines... Figures 2 to 9 A more detailed description of the virtual health status / RUL detector generator 112 is provided.

[0055] The health status detector 114 is a virtual detector and may include one or more machine learning models trained and validated by a virtual health status / RUL detector generator 112. The health status detector 114 may receive sensor data (also referred to as detection data), such as parameter values ​​of vehicle parameters, from the vehicle 104 via data acquisition circuitry 110 or a database. The health status detector 114 may feed the received detection data as input to the machine learning model (or virtual detector), and in response, the machine learning model outputs the health status of components of the vehicle 104, such as an indication of whether the RUL of a vehicle component is less than a predefined threshold. The health status detector 114 may provide or transmit the indication of the health status of vehicle components to the vehicle 104, computer system 106, another remote system or device, or a combination thereof. The following is combined with… Figures 2 to 9 A more detailed description of the health status detector 114 is provided.

[0056] Although Figure 1A single vehicle 104 and a single computer system 106 are shown, but in general, the health status / RUL detection system 102 can be constructed or configured to monitor vehicle components (e.g., one or more of a given type) in multiple vehicles (and / or equipment) 104, and can serve multiple computer systems 106 and / or multiple other remote systems or devices (e.g., reporting health status detection results to these computer systems and / or other remote systems or devices). For example, the health status / RUL detection system 102 or the virtual detector generator 112 can use multiple training datasets to generate multiple health status detectors 114 (or machine learning models), which are used, for example, to detect the RUL or health status of various types of vehicle components. The health status / RUL detection system 102 or the virtual health status / RUL detector generator 112 can generate a separate health status detector 114 (or machine learning model) for each type of vehicle component of vehicle 104. For example, each health status detector 114 can be trained or constructed to detect the RUL or health status of a given type of vehicle component (e.g., an SCR system or a DOC). In some implementations, the health status / RUL detection system 102 can provide the generated health status detector 114 (or machine learning model) to the vehicle 104, computer system 106, or other remote systems or devices. In such implementations, health status / RUL detection can be performed by the vehicle 104, computer system 106, and / or other remote systems or devices.

[0057] Now for reference Figure 2 According to the example embodiment, it is shown Figure 1 A schematic diagram of the controller 200 of the health status / RUL detection system 102 is provided. The controller 200 may represent an example implementation of the health status / RUL detection system 102 or a portion thereof. Additionally, a more detailed block diagram depiction of the vehicle 104 is provided according to an example embodiment.

[0058] The vehicle 104 is shown as including an engine 222, a fuel system 224 coupled to the engine 222, an aftertreatment system 226 coupled to the engine 222, a transmission system 228 coupled to the engine 222, and one or more sensors 230.

[0059] Engine 222 can be any type of engine, such as a gasoline engine, a natural gas engine, a hydrogen engine, a diesel engine, a hybrid engine (e.g., a combination of an internal combustion engine and an electric motor), and / or any other suitable engine. In the depicted example, engine 222 can be a diesel-powered compression ignition engine.

[0060] Aftertreatment system 226 is in exhaust reception communication with engine 222. Aftertreatment system 226 includes a diesel particulate filter (DPF), a diesel oxidation catalyst (DOC), a selective catalytic reduction (SCR) system, and an ammonia slip catalyst (ASC). The DOC is configured to receive exhaust gas from engine 222 and oxidize hydrocarbons and carbon monoxide in the exhaust gas. The DPF is configured to remove particulates, such as soot, from the exhaust gas flowing in the exhaust stream. The DOC may be fluidly coupled to an exhaust duct system to oxidize hydrocarbons and carbon monoxide in the exhaust gas. To properly assist this reduction, the DOC may need to be at a specific operating temperature. In some embodiments, this specific operating temperature is approximately between 200°C and 500°C. In other embodiments, the specific operating temperature is the temperature at which the conversion efficiency of the DOC exceeds a predefined threshold (e.g., the conversion of HC to less harmful compounds, referred to as HC conversion efficiency). The SCR is configured to help reduce NOx emissions by accelerating the NOx reduction process between ammonia and exhaust NOx to diatomic nitrogen and water. If the SCR catalyst is not at or above a certain temperature, the acceleration of the NOx reduction process will be limited, and the SCR may not operate at the regulatory efficiency level. In some embodiments, this specific temperature is approximately 200°C to 600°C. In some implementations, the aftertreatment system 226 may include one or more additional components or fewer components.

[0061] The transmission system 228 is configured to transmit power generated by the engine 222 to the wheels via a mechanical system of gears and gear trains. In other implementations, the final drive unit may be omitted or may be different relative to the wheels mentioned. The transmission system 228 allows the driver to apply power to the vehicle 104 in a controlled manner.

[0062] Sensor 230 may include an engine outlet nitrogen oxide (NOx) sensor, a system outlet NOx sensor, a DEF metering injection sensor, a temperature sensor, a pressure sensor, and / or other sensors to monitor the operating parameters or status of one or more systems or components of vehicle 104.

[0063] The controller 200 includes processing circuitry 202 with a processor 204 and a memory 206. The controller 200 includes one or more data acquisition circuits 110, one or more preprocessing (or data preprocessing) circuits 210, one or more model training circuits 212, one or more model validation (or model testing) circuits 214, one or more machine learning model circuits 216, one or more notification (or output) circuits 218, and a communication interface 220. The data acquisition circuits 110, preprocessing circuits 210, model training circuits 212, model validation circuits 214, machine learning model circuits 216, and / or notification circuits 218 may be part of the processing circuitry 202 or separate components within the controller 200. In some implementations, the preprocessing circuits 210, model training circuits 212, and model validation circuits 214 represent... Figure 1 The circuitry of the virtual RUL detector generator 112 is shown. In some implementations, the machine learning model circuitry 216 and the notification circuitry represent... Figure 1 The circuitry for the health status / RUL detector 114.

[0064] Controller 200 is configured or constructed to receive vehicle training data from one or more vehicles 104 or equipment, computer system 106, and / or other remote systems or devices. The training data may include parameter values ​​from a large set of parameters of vehicle 104. Controller 200 may use the training data to train a machine learning model for detecting the health status and / or relative failure (RUL) of corresponding vehicle components in vehicle 104. Controller 200 may use a first portion of the training data to train at least one machine learning model and a second portion of the training data to validate or test the trained model. Once the machine learning model is generated (e.g., trained and validated), controller 200 may use the model to detect the health status and / or RUL of vehicle components in vehicle 104 and predict "proximate" failure events of the vehicle component based on the detection data received from vehicle 104. Controller 200 may provide or transmit indications of the predicted or detected health status, RUL, and / or failure events of the vehicle component to vehicle 104, computer system 106, and / or other computer systems or devices.

[0065] In some implementations, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 may be embodied or implemented as machine-executable instructions stored in memory 206 (or other memory or storage device) and executed by processor 204. The data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 may be embodied or implemented as a machine or computer-readable medium storing instructions executable by processor 204. As described herein and among other uses, the machine-readable medium also facilitates the performance of certain operations to achieve the reception and transmission of data. For example, the machine-readable medium may provide instructions (e.g., commands, etc.) to, for example, acquire data. In this respect, the machine-readable medium may include programmable logic defining the frequency of data acquisition (or data transmission). The computer-readable medium instructions may include code that can be written in any programming language, including but not limited to Java and any conventional procedural programming language, such as the "C" programming language or similar programming languages. Computer-readable program code can be executed on one processor or multiple remote processors. In the latter scenario, remote processors can be connected to each other via any type of network (e.g., CAN bus, etc.).

[0066] In some implementations, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 can be embodied as hardware units, such as electronic control units. Therefore, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 can be embodied as one or more circuit components, including but not limited to processing circuits, network interfaces, and / or other circuit components. In some implementations, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 can take the form of one or more analog circuits, electronic circuits (e.g., integrated circuits (ICs), discrete circuits, system-on-a-chip (SOC) circuits, microcontrollers, etc.), telecommunications circuits, hybrid circuits, and any other type of "circuit". In this respect, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 may include any type of components to perform or facilitate the implementation of the operations described herein. For example, the circuits described herein may include one or more transistors, logic gates (e.g., NAND, AND, NOR, OR, XOR, NOT, XNOR, etc.), resistors, multiplexers, registers, capacitors, inductors, diodes, wiring, etc. The data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model verification circuit 214, machine learning model circuit 216, and / or notification circuit 218 may also include programmable hardware devices, such as field-programmable gate arrays, programmable array logic, programmable logic devices, etc. The data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model validation circuit 214, machine learning model circuit 216, and / or notification circuit 218 may include one or more memory devices for storing instructions executable by one or more processors of these circuits. The one or more memory devices and processors may have the same definitions provided above regarding memory 206 and processor 204. In some hardware unit configurations, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model validation circuit 214, machine learning model circuit 216, and / or notification circuit 218 may be geographically distributed across separate locations in the cloud. In some implementations, the data acquisition circuit 110, preprocessing circuit 210, model training circuit 212, model validation circuit 214, machine learning model circuit 216, and / or notification circuit 218 may be embodied in a single unit / casing (which is located in...). Figure 2 It is shown as being within or within the controller 200.

[0067] In the illustrated example, controller 200 includes processing circuitry 202 having processor 204 and memory device 206. Processing circuitry 202 may be configured or constructed to execute or implement the instructions, commands, and / or control processes described herein with respect to data acquisition circuitry 110, preprocessing circuitry 210, model training circuitry 212, model verification circuitry 214, machine learning model circuitry 216, and / or notification circuitry 218. The depicted configuration represents circuitry 110 and 210-218 as a machine or computer-readable medium. However, as mentioned above, this illustration is not intended to be limiting, as this disclosure contemplates other embodiments in which circuitry 110 and 210-218, or at least one of circuitry 110 and 210-218, is configured as a hardware unit. All such combinations and variations are intended to fall within the scope of this disclosure.

[0068] Processing circuitry 202 or processor 204 may be constructed or configured to execute or implement the instructions, commands, and / or control processes described herein with respect to data acquisition circuitry 110, preprocessing circuitry 210, model training circuitry 212, model verification circuitry 214, machine learning model circuitry 216, and / or notification circuitry 218. Processor 204 may be implemented as a single-chip or multi-chip processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or any combination thereof designed to perform the functions described herein. Processor 204 may be a microprocessor. Processor 204 may also be implemented as a combination of computing devices, such as a combination of a DSP and a microprocessor, multiple microprocessors, one or more microprocessors combined with a DSP core, or any other such configuration. In some implementations, controller 200 may include one or more processors 204, which may be shared by multiple circuits. For example, data acquisition circuitry 110, preprocessing circuitry 210, model training circuitry 212, model validation circuitry 214, machine learning model circuitry 216, and / or notification circuitry 218 may include or otherwise share the same processor, which in some example embodiments may execute instructions stored or otherwise accessed via different regions of memory. In some implementations, one or more processors may be configured to perform or otherwise perform certain operations independently of one or more coprocessors. Two or more processors may be coupled via a bus to enable independent, parallel, pipelined, or multithreaded instruction execution. All such variations are intended to fall within the scope of this disclosure.

[0069] Memory 206 (e.g., memory cells and / or storage devices) may include one or more devices (e.g., RAM, ROM, flash memory, hard disk storage devices) for storing data and / or computer code used to perform or facilitate the various processes, layers, and modules described herein. Memory 206 may be communicatively connected to processor 204 to provide processor 204 with computer code or instructions for performing at least some of the processes described herein. Furthermore, memory 206 may be or include tangible, non-transitory, or non-volatile memory. Therefore, memory 206 may include database components, object code components, script components, or any other type of information structure to support the various activities and information structures described herein.

[0070] Communication interface 220 may include any combination of wired and / or wireless interfaces (e.g., antennas, transmitters, receivers, transceivers, wire terminals) for data communication with various systems, devices, or networks configured to enable communication with vehicle 104, computer system 106, and / or other remote systems, devices, or databases. For example, communication interface 220 may include Ethernet cards and ports for transmitting and receiving data via Ethernet-based communication networks and / or Wi-Fi transceivers for communication via wireless communication networks. Communication interface 220 may be configured to communicate via local area networks or wide area networks (e.g., the Internet) and may use various communication protocols (e.g., IP, LON, Bluetooth, ZigBee, radio, cellular, near-field communication).

[0071] The controller 200 can communicate with the vehicle 104, computer system 106, and / or other computer systems or devices via communication interface 220 and communication network 108. For example, the controller 200 can communicate with control system 116, telematics unit 118, engine 222, fuel system 224 or components thereof, aftertreatment system 226 or components thereof, transmission system 228, sensors 230, and / or other systems or components of the vehicle 104. In some implementations, the control system 116 of the vehicle 104 monitors and / or manages various systems of the vehicle 104, and the controller 200 communicates with the control system 116 of the vehicle 104 via telematics unit 118 to obtain data about the systems or their components. For example, the control system 116 can collect or record parameter values ​​or measurements from various sensors 230 of the vehicle 104, and the controller 200 can request or receive the collected or recorded parameter values ​​from the control system 116 via telematics unit 118, communication network 108, and / or communication interface 220. In some implementations, controller 200 may command or instruct control system 116 to transmit or provide corresponding parameter values ​​for which parameters. In some implementations, controller 200 may communicate directly with various systems or components of vehicle 104, rather than via control system 116 and / or telematics unit 118.

[0072] Data acquisition circuit 110 is configured or constructed to receive training data, validation data, and / or detection data from a remote system or device, such as vehicle 104, computer system 106, or other computer systems or devices. Data acquisition circuit 110 corresponds to or may be part of data acquisition component 110. Preprocessing circuit 210, model training circuit 212, and model validation circuit 214 are circuits or modules of virtual health status / RUL detector generator 112. Preprocessing circuit 210 is configured or constructed to preprocess training and validation data. Model training circuit 212 is configured or constructed to train a machine learning model for detecting the health status and / or RUL of vehicle components. Model validation circuit 212 is configured or constructed to use validation data to validate or test the machine learning model. Machine learning model circuit 216 and notification circuit 218 may be circuits or modules of health status detector 114. Machine learning model circuit 216 implements a trained (and validated) machine learning model and is configured or constructed to use detection data received from vehicle 104 to determine the health status and / or RUL of vehicle components. Notification circuit 218 is configured or constructed to transmit or provide indications of the detected health status to vehicle 104, computer system 106, and / or other remote systems or devices. (The following is in conjunction with...) Figures 3 to 9 The functionality of these circuits will be described in more detail.

[0073] In some implementations, controller 200 may be the controller of vehicle 104. In such implementations, training and validation of the machine learning model may be performed, for example, by a health status / RUL detection system 102 residing in the cloud. Therefore, data acquisition circuitry 110, preprocessing circuitry 210, model training circuitry 212, and model validation circuitry 214 may be part of the health status / RUL detection system 102. However, machine learning model circuitry 216 and notification circuitry 218 may be implemented within the controller 200 of vehicle 104. After validating the machine learning model (e.g., based on validation results of multiple trained models) and determining the final model for deployment, model validation circuitry 214 can provide the final machine learning model for deployment in the controller of vehicle 104. Therefore, RUL and / or health status detection can be performed in vehicle 104.

[0074] Now for reference Figure 3 According to an example embodiment, a flowchart of a method 300 for generating or training a virtual health status / RUL detector is shown. In short, method 300 may include: a dataset corresponding to various RULs of vehicle components (step 302); and assigning a corresponding health status of a vehicle component to each dataset based on the vehicle component RUL corresponding to the dataset (step 304). Method 300 may include: selecting a subset of parameters from a predefined set of vehicle parameters in the dataset based on corresponding changes that occur in response to changes in health status and / or RUL (step 306). Method 300 may include: generating a machine learning model for detecting the health status and / or RUL of a vehicle component using parameter values ​​from the selected subset of parameters in the dataset (step 308); and providing the machine learning model for detecting the health status and / or RUL of a vehicle component (step 310).

[0075] Now for reference Figures 1 to 3Method 300 may include: one or more processors 204 or data acquisition circuitry 110 obtaining or acquiring a training dataset of vehicle 104 (step 302). One or more processors 204 or data acquisition circuitry 110 may receive or acquire datasets of one or more vehicles 104. Each dataset is recorded from a corresponding vehicle 104 among one or more vehicles under a corresponding RUL of the vehicle component. Each dataset includes parameter values ​​of a predefined set of parameters for vehicle 104. Processor 204 or data acquisition circuitry 110 may receive training data from remote data sources, such as computer system 106, a database, or vehicle 104 and other data sources. For example, computer system 106 may collect parameter values ​​of a parameter set from one or more vehicles 104 over time. In some implementations, vehicles 104 may belong to the same brand and model. In other embodiments, vehicles 104 may belong to different brands and models, but the components and / or systems under study belong to the same brand and model. Computer system 106 may store the collected data in its memory or database, or in a memory or database coupled to the computer system. Computer system 106 can provide or transmit collected data to controller 200 or health status / RUL detection system 102 via communication network 108. Computer system 106 can proactively provide or transmit collected data, or provide or transmit collected data upon request by data acquisition circuit 110. In some implementations, data acquisition circuit 110 can collect data directly from vehicle 104 over time and store the collected data in memory 206 or a database of health status / RUL detection system 102. In some implementations, a cloud database can collect and / or maintain parameter values ​​of a predefined set of parameters for vehicle 104. Data acquisition circuit 110 can retrieve parameter values ​​from the cloud database.

[0076] Now for reference Figure 4 According to an example embodiment, a diagram depicting a process 400 of acquiring or collecting data from vehicle 104 is shown. The control system 116 of vehicle 104 collects data from vehicle sensors 230. In some implementations, the control system 116 may collect a separate dataset for each trip (or for each duration or other operating interval) performed by vehicle 104. As used herein, a trip can be defined as the time interval during which vehicle 104 is continuously turned on. A new trip begins whenever vehicle 104 is turned on (or powered on) and ends when vehicle is turned off (or powered off).

[0077] Each dataset may include parameter values ​​from a predefined set of vehicle parameters collected during the corresponding vehicle trip. The dataset may be referred to as an Embedded Field Performance Analysis (eFPA) dataset and includes the parameter value set for the corresponding trip. The predefined set of vehicle parameters may vary based on the country or state in which vehicle 104 operates. In some implementations, the dataset for each vehicle trip may include parameter values ​​for approximately 450 predefined parameters. In some implementations, the number of predefined vehicle parameters may be approximately 350, approximately 400, or other numbers. Each dataset may include additional information or parameters indicating health and / or performance trends. The size of each dataset may be approximately 2 kilobytes. For a given vehicle 104, control system 116 may collect training datasets for multiple trips over a period of time until at least one failure event of the vehicle component of interest is detected. For example, control system 116 may collect datasets from multiple trips (e.g., over multiple months) while the vehicle component to be monitored (e.g., an SCR system or DOC) is in a healthy state or condition until a failure occurs in the vehicle component. Therefore, the dataset includes at least one dataset collected or recorded during one or more trips while the vehicle component is in a faulty state, and multiple datasets collected or recorded during multiple trips prior to the vehicle component's fault. Training data may include datasets collected or recorded from multiple vehicles 104, wherein multiple datasets are collected or recorded for each vehicle until a fault in the vehicle component of interest is detected or reached. For each vehicle 104, the corresponding dataset includes multiple datasets recorded over a period of time until a fault event or fault state of the vehicle component is detected. At least one dataset from the vehicle's dataset corresponds to a fault event or fault state of the vehicle component.

[0078] The control system 116 can encrypt the recorded dataset and store it in local memory. The control system 116 can transmit or transfer the encrypted dataset to the health status / RUL detection system 102 or a cloud system hosting the health status / RUL detection system 102. Data transmission can be achieved using various communication technologies (such as telematics, EDGE, or mobile applications, such as GUIDANZ MOBILE). When using a mobile application, the mobile device can receive data from the control system 116 via Bluetooth or other wireless technologies. The mobile device can transmit the data received from the control system 116 via a mobile application installed on the mobile device. When using telematics, data can be transmitted or transferred to the health status / RUL detection system 102 via another cloud system 404 (e.g., a cloud system hosting a telematics platform). In some implementations, data collected or recorded by control system 116 can be transmitted to health status / RUL detection system 102 via computing device 406 (e.g., a laptop computer), which receives data from control system 116 via CLIP protocol, for example, and transmits data to health status / RUL detection system 102 via the Internet, for example. Control system 116 can periodically (e.g., daily or weekly, or several times a week) send or transmit the recorded datasets to health status / RUL detection system 102. In some implementations, the controller can store approximately 750 datasets from up to approximately 750 trips, or more datasets. Control system 116 can send or transmit the recorded datasets to health status / RUL detection system 102 in real-time or near real-time, wherein each dataset can be sent to health status / RUL detection system 102 immediately after it is recorded. Training data received from control system 116 can be stored in encrypted form in database 408 of health status / RUL detection system 102.

[0079] The parameter set may include parameters associated with various systems (or components thereof) of vehicle 104. For example, the parameter set may include parameters associated with engine 222, parameters associated with fuel subsystem 224, parameters associated with aftertreatment system 224, parameters associated with transmission system 226, parameters associated with other systems of vehicle 104, or combinations thereof. Training data received or acquired by data acquisition circuitry 110 may include multiple corresponding parameter values ​​across multiple datasets (e.g., corresponding to various vehicle trips) for each parameter in the parameter set. Data sets for a given vehicle 105 are recorded at different time points corresponding to different RULs of the vehicle components of interest (or to be monitored by health status detector 114). Therefore, each vehicle parameter has a corresponding value in the training data corresponding to different RULs of the vehicle components.

[0080] In some implementations, the parameter set may include all or almost all parameters of vehicle 104, such as measurements or values ​​recorded by sensors and / or other components of vehicle 104 for these parameters. For example, the parameter set may be a comprehensive set of parameters for vehicle 104. The parameter set may include parameters associated with or related to one or more specific systems of vehicle 104 (e.g., fuel system 224, aftertreatment system 226, and / or transmission system 228). The health status / RUL detection system 102 and / or controller 200 may initially be unaware of which parameters are related to the detection of the health status and / or RUL of vehicle components.

[0081] Method 300 may include: one or more processors 204 or preprocessing circuitry 210 of computing system 102 assigning a corresponding health state to each vehicle component based on the RUL of the vehicle component corresponding to the dataset (step 304). This assignment of health states can be viewed as labeling the training dataset. In some implementations, the training dataset received by health state / RUL detection system 102 may have already been labeled, wherein each dataset includes an indication of the corresponding health state of the vehicle component. The health state assigned to or mapped to each dataset is defined based on the RUL of the vehicle component when the dataset is recorded in the corresponding vehicle 104.

[0082] Now for reference Figure 5 According to an example embodiment, a diagram 500 is shown depicting how a health status is defined according to RUL. Figure 5 RUL is defined based on the total distance traveled by vehicle 104, for example, as recorded by the odometer. When a vehicle component fails or reaches a failure state, the RUL of the vehicle component equals a predefined value (specifically, zero). If a vehicle component (e.g., an SCR system or POC) fails when the total distance traveled equals a predefined value (denoted as X, which can be expressed in miles or kilometers), then the distance traveled X can be represented as the failure distance (or the vehicle mileage at the point of failure), and the RUL of the vehicle component equals the predefined value (e.g., zero) at the failure distance. In other words, the failure distance (or the vehicle mileage at the point of failure) of a vehicle component represents a reference point for the RUL of the vehicle component.

[0083] like Figure 5As depicted, the Range of Usage (RUL) of a vehicle component prior to the point of failure is defined as the difference between the distance traveled by the vehicle 104 to the failure point X and the current total distance traveled. For example, RUL can be defined or calculated as RUL = Failure Mileage (km) – Travel Mileage (km), where travel mileage (km) represents the total distance traveled by the vehicle to the corresponding travel distance or the distance provided by the vehicle's odometer for that travel distance. For example, if the current total distance traveled by vehicle 104 (e.g., as recorded by the odometer) is 10,000 km less than the failure distance X, then the RUL of the vehicle component is 10,000 km (km). In other words, RUL represents the additional distance the vehicle must travel or travel before the vehicle component fails or reaches a failure state.

[0084] For example, a training dataset of vehicle 104 recorded over multiple trips of vehicle 104, each dataset may include an indication of the total distance traveled at the time of recording (e.g., distance traveled as read from an odometer). Datasets corresponding to failures of vehicle components (e.g., datasets corresponding to trips when a vehicle component fails or reaches a failure state) may be indicated or labeled as failure datasets. One or more processors 204 or preprocessing circuitry 210 may determine the RUL corresponding to each dataset as the difference between the distance traveled associated with the failure dataset and the distance traveled in the dataset.

[0085] One or more processors 204 or preprocessing circuits 210 can determine or set thresholds for RUL to indicate an impending fault event. For example, Figure 5 The threshold is 10,000 km. When the RUL of a vehicle component is less than (or less than or equal to) the RUL threshold, one or more processors 204 or preprocessing circuits 210 can define or determine the health status of the vehicle component as "unhealthy" or "unhealthy state". When the RUL is greater than the RUL threshold, one or more processors 204 or preprocessing circuits 210 can determine or define the health status of the vehicle component as "healthy" or "healthy state". One or more processors 204 or preprocessing circuits 210 can assign (or label each dataset recorded before the vehicle component fails) a "healthy state" or "unhealthy state" based on the corresponding RUL. Labeling the dataset with the health status of the vehicle component allows training a machine learning model for predicting the health statistics of the vehicle component.

[0086] In some implementations, RUL can be defined as the remaining operating time of a vehicle component before it fails or reaches its failure state. Operating time can be defined as the time during which the vehicle 104 or equipment is activated or the vehicle is moved. In other words, RUL can be defined using units of time rather than units of distance.

[0087] One or more processors 204 or preprocessing circuits 210 can preprocess the dataset to estimate, fill in, or supplement missing data points. For example, for some vehicle parameters, if the parameter or corresponding event is not triggered, the corresponding parameter value will not be recorded. For example, if no corresponding fault is triggered in the afterprocessing system 226, the dataset will not include derating time. In this case, one or more processors 204 or preprocessing circuits 210 can assign predefined values ​​(e.g., zero values) to parameters indicating derating time. Generally, one or more processors 204 or preprocessing circuits 210 can clean the dataset so that all parameters have corresponding values ​​in all datasets. In some implementations, one or more processors 204 or preprocessing circuits 210 can remove datasets that are determined to be corrupted (e.g., missing a relatively large number of parameter values ​​or exhibiting unreasonable parameter values) from the training data.

[0088] Method 300 may include: one or more processors 204 or preprocessing circuitry 210 selecting a subset of parameters from a predefined set of parameters in a dataset based on corresponding changes that occur in response to changes in the health status and / or RUL of a vehicle component (step 306). In selecting the parameter subset, one or more processors 204 or preprocessing circuitry 210 may identify or determine parameters that exhibit higher sensitivity to changes in the health status and / or RUL of a vehicle component. For example, vehicle parameters exhibiting a relatively high range of variation in response to changes in the health status and / or RUL of a vehicle component are better predictors of the health status and / or RUL of the vehicle component.

[0089] The preprocessing circuit 210 may initially filter the maintained training data based on at least one of the attributes of geographic location, vehicle brand and / or model and / or one or more other characteristics or data of the vehicle. This initial filtering step may depend on the characteristics of the training data available to the health status / RUL detection system. For example, depending on the characteristics or span of the training data maintained and / or provided by the data source, the initial filtering step may be optional or omitted.

[0090] Preprocessing circuit 210 can filter parameters based on cross-correlation or multicollinearity between different parameters of vehicle 104. Preprocessing circuit 210 can reduce the number of parameters to be considered by eliminating redundancy associated with substantially correlated or collinear parameters of vehicle 104. Preprocessing circuit 210 can calculate a cross-correlation index (e.g., correlation coefficient) between sequences of parameter values ​​of different parameters of vehicle 104. Preprocessing circuit 210 can maintain a subset of parameters (or corresponding sequences of parameter values) of vehicle 104 with relatively low correlation coefficients (e.g., below a predefined threshold). For example, if the correlation coefficient of a pair of parameters or corresponding sequences of parameter values ​​is higher than a predefined threshold, preprocessing circuit 210 can eliminate one of these parameters or corresponding sequences of parameter values. The predefined threshold for the correlation coefficient can be a predefined value, such as 0.9, 0.85, or 0.8, and other possible values. Applying filtering based on cross-correlation or multicollinearity between different parameters reduces the number of parameters (or corresponding sequences or parameter values) from 365 to 205.

[0091] As discussed above, training data can be labeled by dataset (or by trip). For example, each dataset (except for datasets associated with failure events) may include indicators indicating whether the dataset is associated with a "healthy state" or an "unhealthy state". Thus, for a given vehicle parameter, corresponding values ​​across multiple datasets are associated with the corresponding health state and / or corresponding RUL of the vehicle component. Preprocessing circuit 210 can evaluate the variation or range of variation of each parameter as a function of the health state and / or RUL of the vehicle component of interest. Preprocessing circuit 210 can filter the parameters of vehicle 104 based on the variation or range of the corresponding values ​​as a function of the RUL and / or health state of the vehicle component (or in response to a change in the RUL and / or health state). For a given parameter, if the corresponding value exhibits a relatively significant or substantial change with changes in the health state and / or RUL of the vehicle component, it indicates that the parameter depends on or serves as a relatively good predictor of the health state and / or RUL of the vehicle component.

[0092] Preprocessing circuit 210 can select a subset of parameters from a predefined set of parameters of vehicle 104 as parameters with the largest range of variation in response to changes in the relative fitness level (RUL) and / or health status of vehicle components. Preprocessing circuit 210 can rank and / or classify the vehicle parameters based on the relationship between the corresponding change or range of variation and changes in the RUL and / or health status of the vehicle components. In some implementations, preprocessing circuit 210 can use an unsupervised machine learning model to rank and / or classify the parameters of vehicle 104 based on the corresponding change or range of variation that occurs in response to changes in the RUL and / or health status of vehicle components. In some implementations, preprocessing circuit 210 can use an extreme gradient boosting (XGBoost) algorithm to select the subset of parameters with the largest range of variation or the greatest impact on the RUL and / or health status of vehicle components in response to changes in the RUL and / or health status of vehicle components. Specifically, preprocessing circuit 210 may use the XGBoost algorithm to classify and / or rank parameters based on corresponding changes or ranges of change (in response to changes in the RUL and / or health status of vehicle components) and / or based on the corresponding impact on the RUL and / or health status of vehicle components, and select a set of parameters with the largest range of change or the largest impact. In some implementations, preprocessing circuit 210 may use other unsupervised machine learning models (e.g., models other than the XGBoost algorithm), other ranking or classification algorithms (e.g., not based on machine learning models), or other techniques to identify parameters that exhibit relatively high variability in response to changes in the RUL and / or health status of vehicle components. By applying the XGBoost algorithm to the parameter values ​​in the training dataset, preprocessing circuit 210 may identify a relatively small number of relevant parameters compared to the total number of parameters in a predefined set.

[0093] Now for reference Figure 6 According to an example embodiment, a graph 600 is shown depicting the relationship between Shapley Explanation (SHAP) values ​​of multiple selected vehicle parameters and the health status of the DOC. Specifically, Figure 6 The distribution of SHAP values ​​for 20 selected parameters of vehicle 104, derived from field data, is illustrated. The left side of the graph or chart represents the SHAP values ​​corresponding to the health status of the DOC, while the right side depicts the SHAP values ​​corresponding to the unhealthy status of the DOC. The range of variation of the SHAP value of a given parameter with changes in the health status of the DOC is a good indicator of the effectiveness of that parameter in detecting the health status of the DOC. The preprocessing circuit 210 can select a subset of parameters based on the corresponding SHAP values ​​to train a machine learning model for detecting the health status of the DOC.

[0094] Table 1 below describes Figure 6The list shows a ranking of 18 of the 20 parameters, along with their corresponding descriptions. This ranking indicates the order of parameter effectiveness in detecting the health status of the DOC. For example, the Minimum Model Basis Soot Load (MBSLE) parameter in post-processing is the most predictive parameter for the health status of the DOC (based on SHAP values), the Maximum Combined Soot Load (CSLE) parameter in post-processing is the second most predictive parameter, the Maximum MBSLE is the third most predictive parameter, the Average Differential Soot Load Estimate (DPSLE) offset adjustment parameter used to measure ash load adjustment is the fourth most predictive parameter, and the Maximum Urea Pump Pressure (related to DEF pressure) is the fifth most predictive parameter. In some implementations, the selected parameter set may include at least one of these five parameters. Table 1. Parameters used to predict DOC health status and / or DOC RUL.

[0095] According to the example embodiment, it is shown Figure 7 Graph 620 shows the relationship between SHAP values ​​of several selected vehicle parameters and the health status of the SCR system. Specifically, Figure 7 The distribution of SHAP values ​​for 20 selected parameters of vehicle 104 and their relationship to the health status of the SCR system are illustrated. The left side of the graph or chart represents the SHAP values ​​corresponding to the health status of the SCR system, while the right side depicts the SHAP values ​​corresponding to the unhealthy status of the SCR system. The range of variation of the SHAP value of a given parameter with changes in the health status of the SCR system is a good indicator of the effectiveness of that parameter in detecting the health status of the SCR system. The preprocessing circuit 210 can select a subset of parameters based on the corresponding SHAP values ​​to train a machine learning model for detecting the health status of the SCR system.

[0096] Table 2 below describes Figure 7 The list shows a ranking of 19 of the 20 parameters, along with their corresponding descriptions. This ranking indicates the order of parameter effectiveness in detecting the health status of the SCR. For example, the maximum CSLE parameter is the most predictive parameter of SCR health status (based on the SHAP value), the maximum MBSLE is the second most predictive parameter, the parameter representing the average urea pump command (UL2 system) when the pressure is in closed loop but without metering injection is the third most predictive parameter, the maximum diesel particulate filter (DPF) differential pressure parameter is the fourth most predictive parameter, and the parameter representing the average compressor inlet pressure when the equipment (or vehicle) is operating is the fifth most predictive parameter. In some implementations, the selected parameter set may include at least one of these five parameters. Table 2. Parameters used to predict SCR health status and / or SCR RUL.

[0097] According to the example embodiment, it is shown Figure 8 A graph 620 illustrates the relationship between SHAP values ​​of several selected vehicle parameters and the health status and / or RUL of a filter (such as a specific filter, e.g., a diesel particulate filter). In this example, the filter is a DPF. Specifically, Figure 8 The distribution of SHAP values ​​for 20 selected parameters of vehicle 104 and their relationship to the health status and / or RUL of the DPF are illustrated. The left side of the graph or chart represents the SHAP values ​​corresponding to the health status of the filter, while the right side depicts the SHAP values ​​corresponding to the unhealthy status of the filter. The range of variation of the SHAP value of a given parameter with changes in the health status of the SCR system can be a good indicator of the effectiveness of that parameter in detecting the health status of the DPF. The preprocessing circuit 210 can select a subset of parameters based on the corresponding SHAP values ​​to train a machine learning model for detecting the health status of the DPF.

[0098] Table 3 below describes Figure 8 The list shows a ranking of 20 parameters and their corresponding descriptions. This ranking indicates the order of effectiveness of the parameters in detecting filter health and / or RUL. For example, in some cases, the Minimum Model Basis Soot Load (MBSLEMIN) parameter may be the most predictive parameter for filter health and / or RUL (based on SHAP value), the Maximum Pressure Change of Particulate Filter (DPFDPMAX) parameter may be the second most predictive parameter, the Maximum Pressure Change Soot Load (DPSLEMAX) parameter may be the third most predictive parameter, the Average Model Basis Soot Load (MBSLEMEAN) parameter may be the fourth most predictive parameter, and the Maximum Combined Soot Load (CSLEMAX) parameter may be the fifth most predictive parameter. In some implementations, the selected parameter set may include at least one of these five parameters. Table 3. Parameters used to predict filter health status and / or filter RUL.

[0099] This filtering of parameters based on corresponding changes or corresponding SHAP values ​​allows for the identification of parameters of vehicle 104 that will be most effective in detecting the health status of vehicle components of interest. Similarly, the reduced number of parameters required for health status detection makes the training computation of the virtual health status / RUL detector 114 more efficient, simplifies the detector, and significantly reduces the amount of data to be transmitted to the health status / RUL detection system 102 or controller 200. For example, during the detection phase, only parameter values ​​corresponding to selected parameters are transmitted to the controller 200 or health status / RUL detection system 102.

[0100] It should be noted that the preprocessing circuit 210 can perform any combination of the data preprocessing steps described above. The preprocessing circuit 210 can perform the combined data preprocessing steps in any order. In other words, the order of the data preprocessing steps described above is not restrictive, but rather represents an exemplary preprocessing order as well as other possible orders. Similarly, it should be noted that the parameters of vehicle 104 are referred to herein as characteristics of vehicle 104 or its subsystems.

[0101] Return to reference Figures 1 to 3 Method 300 may include: processor 204 or model training circuit 212 using parameter values ​​from a selected subset of parameters to generate a machine learning model for health status and / or RUL detection (step 308). Generation of the machine learning model may include: processor 204 or model training circuit 212 training and validating at least one machine learning model. In some implementations, a first portion of the parameter values ​​from the selected parameters may be used to train the machine learning model by model training circuit 212, and a second portion may be used to validate the machine learning model by model validation circuit 214. In some other implementations, validation data may be obtained from another dataset (e.g., a validation dataset) based on a subset of parameters identified or selected by preprocessing circuit 210.

[0102] Model training circuit 212 can train one or more machine learning models using parameter values ​​from a subset of parameters identified or selected by preprocessing circuit 210. Machine learning models may include random forest models, statistical learning models, neural networks (e.g., deep neural networks), and / or another type of machine learning model. Training a machine learning model may include: model training circuit 212 selecting initial coefficients or parameters for the model; model training circuit 212 selecting the model's architecture or configuration and assigning initial values ​​to the parameters of the model to be estimated; and model training circuit 212 applying the selected subset of parameters (e.g., ...). Figure 6 , Figure 7 or Figure 8The parameter values ​​of the 20 parameters (as illustrated in the example) are used as input to the machine learning model, and the output of the machine learning model (e.g., the predicted health status of vehicle components) is compared with the parameter values ​​of the selected parameters or the labels or health status associated with the corresponding dataset.

[0103] For example, a first set of parameter values ​​from a selected subset can be associated with a dataset corresponding to the health status of a vehicle component of interest. When the first set of parameter values ​​from the selected subset is used as input to a machine learning model, the model training circuit 212 can compare the model's output with the health status. Similarly, a second set of parameter values ​​from a selected subset can be associated with a dataset corresponding to the health status of a vehicle component. When the second set of parameter values ​​from the selected subset is used as input to a machine learning model, the model training circuit 212 can compare the model's output with an unhealthy status. Note that the machine learning model is constructed or configured to provide an estimate of the health status in response to a set of parameter values ​​from the selected subset provided as input.

[0104] If the output of the machine learning model does not match the health state associated with the input value, the model training circuit 212 can update or modify one or more coefficients or parameters of the machine learning model. The model training circuit 212 can update or modify the coefficients or parameters of the machine learning model individually after each mismatched output, or all at once after feeding all available parameter value sets of a selected subset of parameters to vehicle 104. The model training circuit 212 can iteratively feed the parameter value sets of the selected subset of parameters to vehicle 104 and update the coefficients or parameters of the machine learning model until, for example, all outputs of the machine learning model match the corresponding label or health state associated with the input data. In some implementations, the model training circuit 212 can iteratively feed the parameter value sets of the selected subset of parameters to vehicle 104 and update the coefficients or parameters of the machine learning model until some other convergence criterion(s) is satisfied.

[0105] In some implementations, the model training circuit 212 can use parameter values ​​from a selected subset of parameters to train multiple machine learning models for health status detection. For example, the model training circuit 212 can use parameter values ​​from a selected subset of parameters to train, for example, multiple random forest models (or random decision forests) with different architectures and / or different total numbers of nodes. Training multiple machine learning models provides multiple trained models to choose from, for example, during the validation phase.

[0106] Model validation circuit 214 can use validation or test data to validate the trained machine learning model provided by model training circuit 212. Validation or test data includes parameter values ​​of a subset of parameters selected by preprocessing circuit 210. The validation or test data may include some groups of parameter values ​​selected by preprocessing circuit 210 that are associated with the health status of vehicle components, and other groups that are associated with the unhealthy status of vehicle components. The validation or test data differs from the data used to train the machine learning model. For example, the parameter values ​​of the selected subset of parameters used for validation may be associated with different vehicle trips, compared to the parameter values ​​of the subset of parameters used to train the machine learning model. In some implementations, the parameter values ​​of the selected subset of parameters used for validation may be recorded at different time points and / or by different vehicles, compared to the parameter values ​​of the selected subset of parameters used to train the machine learning model.

[0107] Validating or testing a trained machine learning model may include: model validation model 214 feeding various groups of parameter values ​​(e.g., groups of 20 parameter values ​​corresponding to 20 selected parameters) of a subset of parameters selected by preprocessing circuitry 210, and determining whether the output of each group matches the health status of the corresponding dataset. The validation or testing process evaluates the reliability of the trained machine learning model in detecting the health status of vehicle components based on an input dataset different from the training dataset. In some implementations, model validation model 214 may validate or test each of multiple trained machine learning models and select the best-performing model for deployment to detect the health status of the vehicle components of interest.

[0108] Method 300 may include: processor 204 or model validation circuit 214 providing a machine learning model for detecting the health status of a vehicle component of interest (step 310). For example, processor 204 or model validation circuit 214 may provide a validated model for deployment (or a model selected based on the validation results of multiple trained models) to detect the health status of a vehicle component, as described below. Figure 9 As described.

[0109] While method 300 is primarily described regarding the training and deployment of a machine learning model or virtual detector for detecting the health status of vehicle components, the same method can also be used to train a machine learning model for estimating the RUL of vehicle components. In some implementations, the health status / RUL detection system 102 or controller 200 may train a machine learning model to estimate the RUL, rather than detecting the health status of vehicle components or as a supplement to detecting the health status of vehicle components. The machine learning model may be trained to provide an estimated RUL of the vehicle component as output, and the model training circuit 212 may adjust the model's coefficients and / or parameters based on the match or difference between the model's output during training and RUL values ​​associated with different datasets. The preprocessing circuit 210 may select a subset of parameters based on changes in a predefined set of parameters in response to changes in RUL. In some implementations, the health status / RUL detection system 102 or controller 200 may train a machine learning model to detect or estimate the RUL and health status of vehicle components.

[0110] Method 300 can be used to train machine learning models for detecting or estimating the RUL and / or health status of various types of vehicle components or systems, such as components of the aftertreatment system and / or other systems or components of vehicle 104. In some implementations, the health status / RUL detection system 102 or controller 200 can train or generate multiple machine learning models for detecting or estimating the RUL and / or health status of various vehicle components. For example, the health status / RUL detection system 102 or controller 200 can train or generate a first machine learning model for detecting or estimating the RUL and / or health status of the DOC and a second machine learning model for detecting or estimating the RUL and / or health status of the SCR system.

[0111] Now for reference Figure 9 According to the example embodiment, the use of Figure 3 The flowchart illustrates a method 800 for detecting the health status of vehicle 104 and / or equipment components and / or systems using a virtual health status / RUL detector (or machine learning model) generated in the process. In short, method 800 may include: acquiring or receiving parameter values ​​of multiple parameters of vehicle (or equipment) 104 (step 802); using the machine learning model and the parameter values ​​of the multiple parameters of vehicle (or equipment) 104 to determine the health status and / or RUL of vehicle (or equipment) 104 (step 804). Method 800 may also include: providing indications of the health status and / or RUL for display on a user interface (step 806).

[0112] Now for reference Figure 1 , Figure 2 and Figure 9Method 800 may include: processor 204 or data acquisition circuit 110 acquiring or receiving parameter values ​​of multiple parameters of vehicle (or equipment) 104 (step 802). These multiple parameters may be a subset of parameters of the selected or identified vehicle, as described above in combination with preprocessing circuit 210, step 306, and... Figure 6 , Figure 7 and / or Figure 8 As described. In other words, multiple parameters are identified or selected during the development or training phase of the machine learning model. In some embodiments, multiple parameters are associated with vehicle components such as diesel oxidation catalysts (DOC), selective catalytic reduction (SCR) systems, or filters such as diesel particulate filters (DPF). In one example arrangement, when multiple parameters are associated with the DOC, the subset of parameters includes at least one of the following: Minimum Model Basis Soot Load (MBSLE) parameter in the aftertreatment, Maximum Combined Soot Load (CSLE) parameter in the aftertreatment, Maximum MBSLE parameter, parameter representing the offset adjustment of the Mean Differential Soot Load Estimate (DPSLE), or Maximum Urea Pump Pressure parameter. In another example, when multiple parameters are associated with SCR, the subset of parameters includes at least one of the following: the maximum combined soot load in the aftertreatment (CSLE) parameter, the maximum model-based soot load in the aftertreatment (MBSLE) parameter, the parameter representing the average urea pump command when the pressure is in the closed loop but there is no metering injection, the maximum diesel particulate filter (DPF) differential pressure parameter, or the parameter representing the average compressor inlet pressure when the equipment is operating.

[0113] In some implementations, data acquisition circuit 110 or virtual detector generator 112 can notify vehicle 104 of selected or identified parameters, and vehicle 104 can transmit recorded parameter values ​​of a subset of identified or selected parameters to machine learning model circuit 216. Data acquisition circuit 110 or machine learning model circuit 216 can specify multiple identified or selected parameters in a request for recorded parameter values ​​to detect fuel contamination. Parameter values ​​are recorded by sensor 230 or other components of vehicle 104 during the journey of vehicle 104.

[0114] In some implementations, the control system 116 of the vehicle 106 can transmit, for example... Figure 4 The parameter values ​​discussed herein. The control system 116 can collect parameter values ​​of multiple parameters for each trip and transmit the parameter values ​​to the health status / RUL detection system 102. In some implementations, the controller 116 can collect a dataset of parameter values ​​for vehicle trips performed each day and transmit the entire dataset collected for the day at once.

[0115] Method 800 may include: processor 204 or machine learning model circuitry 216 using a machine learning model and parameter values ​​of multiple parameters of vehicle 104 to determine the health status and / or RUL of vehicle components (step 804). The machine learning model may be a model provided by model validation circuitry 214 and is configured or constructed to provide or output an estimate of the health status and / or RUL of vehicle components when fed parameter values ​​of multiple parameters as input.

[0116] Method 800 may include: processor 204 or notification circuit 218 providing or transmitting to a remote computer system an indication of the health status and / or relative health condition (RUL) of vehicle components detected or estimated by machine learning model circuit 216 for display on a user interface. The remote computer system may include a computer system of vehicle 104 or computer system 106. When transmitted to vehicle 104, the indication may cause control system 116 to display a signal indicating the detected health status and / or RUL on the dashboard of vehicle 104. When transmitted to computer system 106, if the indication indicates an unhealthy state or a relatively low RUL (e.g., less than a predefined threshold), the indication may be displayed on the user interface and / or may trigger a vehicle maintenance schedule.

[0117] In some implementations, method 800 may include: processor 204 or notification circuit 218 providing or transmitting instructions to vehicle 104 or control system 116 to automatically trigger events in response to the detection of an unhealthy state of a vehicle component. For example, the instructions (or commands) may cause control system 116 to activate or power a system or device (such as a heater) of vehicle 104, trigger an error code associated with the detected unhealthy vehicle component, restrict the vehicle to one or more driving or operating modes, impose a driving speed limit, and / or other events (e.g., derating). Such events or actions may be triggered until vehicle 104 or equipment is serviced and the unhealthy component is replaced or repaired. In some implementations, the instructions (or commands) may cause control system 116 to display a warning signal on the dashboard of vehicle 104 until vehicle 104 or equipment is serviced and the unhealthy component is replaced or repaired.

[0118] refer to Figure 3 and Figure 8Controller 200 (and / or one or more of its components, such as processing circuitry 202) is configured to use at least one first machine learning model to identify a subset of parameters related to the health status and / or relative health and safety (RUL) of a component (e.g., DOC, SCR system, DPF, etc.), as described herein with respect to method 300, and to use at least one second machine learning model to determine the health status and / or RUL of the component, as described herein with respect to method 800. The first machine learning model may be a classifier model (e.g., XGBoost, etc.) used to classify parameters and identify the most relevant parameters (e.g., based on SHAP values, etc.), as described herein with respect to method 300. Controller 200 then generates a second machine learning model based on the relevant parameters, as described herein with respect to method 300. The second machine learning model may be a random forest model, a statistical learning model, a neural network (e.g., a deep neural network), and / or another type of machine learning model, such as a regression model. Controller 200 uses the second machine learning model to determine the health status and / or RUL of the component, as described herein with respect to method 800. Therefore, controller 200 uses at least two different machine learning models to determine the health status and / or RUL of the components.

[0119] As used herein, the terms “about,” “approximately,” “substantially,” and similar terms are intended to have a broad meaning consistent with common and accepted usage by one of ordinary skill in the art to which the subject matter of this disclosure pertains. Those skilled in the art who read this disclosure will understand that these terms are intended to allow for the description of certain features described and claimed, without limiting the scope of these features to the precise numerical ranges provided. Therefore, these terms should be interpreted as indicating that non-substantial or irrelevant modifications or alterations to the described and claimed subject matter are considered to be within the scope of this disclosure as set forth in the appended claims.

[0120] It should be noted that the term “exemplary” and its variations, as used herein to describe various embodiments, are intended to indicate that such embodiments are possible examples, representations or illustrations of possible embodiments (and such terms are not intended to imply that such embodiments are necessarily extraordinary or best examples).

[0121] As used herein, the term “coupling” and its variations mean that two components are directly or indirectly connected to each other. This connection can be stationary (e.g., permanent or fixed) or movable (e.g., removable or releasable). Such a connection can be achieved when two components are directly coupled to each other, when two components are coupled to each other using one or more separate intermediate components, or when two components are coupled to each other using an intermediate component that is integral with one of the two components to form a single whole. If “coupling” or its variations are modified by an additional term (e.g., direct coupling), the general definition of “coupling” provided above is modified by the common linguistic meaning of the additional term (e.g., “direct coupling” means joining two components without any separate intermediate component), resulting in a narrower definition than the general definition of “coupling” provided above. Such coupling can be mechanical, electrical, or fluid. For example, circuit A being communicatively “coupled” to circuit B can mean that circuit A communicates directly with circuit B (i.e., without intermediaries) or indirectly with circuit B (e.g., through one or more intermediaries).

[0122] References to the location of elements herein (e.g., “top,” “bottom,” “above,” “below”) are used only to describe the orientation of the various elements in the accompanying drawings. It should be noted that the orientation of the various elements may differ according to other exemplary embodiments, and such variations are intended to be covered by this disclosure.

[0123] Although Figure 2 Various circuits with specific functionalities are illustrated herein; however, it should be understood that controller 140 may include any number of circuits for performing the functions described herein. For example, the activities and functionalities of offset circuit 110, gain circuit 210, reducing agent delivery circuit 212, response hysteresis circuit 214, degradation level circuit 216, diagnostic threshold circuit 218, and / or diagnostic decision circuit 220 may be combined in multiple circuits or combined as a single circuit. Additional circuits with additional functionalities may also be included. Furthermore, controller 140 may further control other activities beyond the scope of this disclosure.

[0124] As mentioned above, and in one configuration, the "circuit" can be implemented in a machine-readable medium for use with various types of processors (such as...). Figure 2The executable code is executed by the processor 204. Executable code may, for example, comprise one or more physical or logical blocks of computer instructions, which may be organized, for example, into objects, procedures, or functions. However, an executable program need not be physically located together, but may comprise different instructions stored in different locations, which, when logically combined, comprise circuitry and implement the intended purpose of the circuitry. In practice, the circuitry of computer-readable program code may be a single instruction or multiple instructions, and may even be distributed across several different code segments, different programs, and several memory devices. Similarly, runtime data may be identified and exemplified within the circuitry herein, and may be embodied in any suitable form and organized within any suitable type of data structure. This runtime data may be collected as a single dataset, or may be distributed across different locations (including different storage devices), and may exist at least partially as electronic signals on a system or network.

[0125] While the term "processor" has been briefly defined above, the terms "processor" and "processing circuitry" should be interpreted broadly. In this respect, and as mentioned above, a "processor" can be implemented as one or more processors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), or other suitable electronic data processing components configured to execute instructions provided by memory. One or more processors can be in the form of a single-core processor, a multi-core processor (e.g., a dual-core processor, a triple-core processor, a quad-core processor, etc.), a microprocessor, etc. In some embodiments, one or more processors can be located externally to the device; for example, one or more processors can be remote processors (e.g., cloud-based processors). Alternatively or additionally, one or more processors can be internal to the device and / or local. In this respect, a given circuitry or its components can be locally configured (e.g., as part of a local server, a local computing system, etc.) or remotely configured (e.g., as part of a remote server, such as a cloud-based server). For this purpose, a "circuitry" as described herein can include components distributed across one or more locations.

[0126] Although the accompanying drawings and descriptions may illustrate a specific order of method steps, the order of these steps may differ from the order depicted and described unless otherwise specified above. Similarly, unless otherwise stated above, two or more steps may be performed simultaneously or partially simultaneously. Such variations may depend, for example, on the chosen software and hardware system and the designer's choices. All such variations are within the scope of this disclosure. Likewise, the software implementation of the described methods can be achieved using standard programming techniques with rule-based logic and other logic to accomplish various connection steps, processing steps, comparison steps, and decision steps.

Claims

1. A system for training a machine learning model for detecting the health status of vehicle components, the system comprising: One or more processors; and The memory stores executable instructions that, when executed by the one or more processors, cause the one or more processors to: Obtain multiple datasets for one or more vehicles, each dataset being recorded from a corresponding vehicle among the one or more vehicles under a corresponding remaining useful life (RUL) of a vehicle component, and including parameter values ​​from a predefined set of parameters for the vehicle. Based on the corresponding RUL of the vehicle component, assign the corresponding health status of the vehicle component to each of the plurality of datasets; Based on the changes that occur in response to changes in the health status of the vehicle components, the multiple datasets are used to select a subset of parameters from the predefined parameter set; A machine learning model for detecting the health status of the vehicle components is trained using parameter values ​​from the subset of parameters in the plurality of datasets. as well as A machine learning model is provided for detecting the health status of the vehicle components.

2. The system of claim 1, wherein the vehicle component is a diesel oxidation catalyst (DOC).

3. The system of claim 2, wherein the subset of parameters includes at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in post-processing; The parameter for maximum combined soot load (CSLE) in post-processing; Maximum MBSLE parameter; The parameter representing the offset adjustment of the mean pressure differential soot loading estimate (DPSLE); or Maximum urea pump pressure parameters.

4. The system of claim 1, wherein the vehicle component is a selective catalytic reduction (SCR) system.

5. The system of claim 4, wherein the subset of parameters includes at least one of the following: The parameter for maximum combined soot load (CSLE) in post-processing; Maximum Model Base Soot Load (MBSLE) parameter in post-processing; The parameter representing the average urea pump command when the pressure is in the closed loop but there is no metering injection; Maximum diesel particulate filter (DPF) differential pressure parameter; or This parameter represents the average compressor inlet pressure when the equipment is in operation.

6. The system of claim 1, wherein the vehicle component is a particulate filter.

7. The system of claim 6, wherein the subset of parameters includes at least one of the following: Minimum Model Base Soot Load (MBSLEMIN) parameter; Maximum pressure variation (DPFDPMAX) parameter for particulate filters; Maximum pressure variation soot load (DPSLEMAX) parameter; Average model baseline soot load (MBSLEMEAN) parameter; or Maximum combined soot load (CSLEMAX) parameter.

8. The system according to claim 1, wherein: The machine learning model is a second machine learning model; and The selection of the subset of parameters from the predefined parameter set is based on the use of a first machine learning model, different from the second machine learning model, to identify the changes in the predefined parameter set in response to changes in the health status of the vehicle components.

9. The system of claim 8, wherein the first machine learning model is a classifier model.

10. The system of claim 1, wherein the machine learning model comprises at least one of the following: Random forest model; Statistical learning models; or Neural network.

11. The system of claim 1, wherein the health status of the vehicle component includes an unhealthy status based on the vehicle component's RUL being less than a predefined threshold and a healthy status based on the vehicle component's RUL being greater than the predefined threshold.

12. The system of claim 11, wherein the predefined threshold is defined as the additional distance that the vehicle corresponding to the vehicle component must travel before the performance of the vehicle component reaches a fault state.

13. A method for training a machine learning model for detecting the health status of vehicle components, the method comprising: The computer system obtains multiple datasets for one or more vehicles, each dataset being recorded from a corresponding vehicle among the one or more vehicles under the corresponding remaining useful life (RUL) of a vehicle component, and including parameter values ​​of a predefined set of parameters for the vehicle. The computer system assigns a corresponding health status of the vehicle component to each of the plurality of datasets based on the corresponding RUL of the vehicle component; The computer system selects a subset of parameters from the predefined parameter set based on changes in the health status of the vehicle components, using the multiple datasets. The computer system uses parameter values ​​from the subset of parameters in the plurality of datasets to train a machine learning model for detecting the health status of the vehicle components. as well as The computer system provides the machine learning model for detecting the health status of the vehicle components.

14. The method of claim 13, wherein the vehicle component is at least one of a diesel oxidation catalyst (DOC), a selective catalytic reduction (SCR) system, or a diesel particulate filter (DPF).

15. The method of claim 14, wherein the subset of parameters includes at least one of the following: Minimum Model Base Soot Load (MBSLE) parameter in post-processing; The parameter for maximum combined soot load (CSLE) in post-processing; Maximum MBSLE parameter; The parameter represents the offset adjustment of the differential pressure carbon load estimate (DPSLE); Maximum urea pump pressure parameters Maximum Model Base Soot Load (MBSLE) parameter in post-processing; The parameter representing the average urea pump command when the pressure is in the closed loop but there is no metering injection; Maximum diesel particulate filter (DPF) differential pressure parameter; A parameter representing the average compressor inlet pressure during equipment operation. Maximum pressure variation soot load (DPSLEMAX) parameter; Average model baseline soot load (MBSLEMEAN) parameter; or Maximum combined soot load (CSLEMAX) parameter.

16. The method of claim 13, wherein The machine learning model is a second machine learning model; and The selection of the subset of parameters from the predefined parameter set is based on the use of a first machine learning model, different from the second machine learning model, to identify the changes in the predefined parameter set in response to changes in the health status of the vehicle components.

17. The method of claim 16, wherein the first machine learning model is a classifier model.

18. The method of claim 11, wherein the machine learning model comprises at least one of the following: Random forest model; Statistical learning models; or Neural network.

19. The method of claim 11, wherein the vehicle component is defined as being in the unhealthy state based on the RUL of the vehicle component being less than a predefined threshold.

20. A non-transitory computer-readable medium storing instructions that, when executed by one or more processors of at least one processing circuit, cause the at least one processing circuit to perform operations, the operations comprising: Obtain a dataset of parameter values ​​for the vehicle at the corresponding Remaining Useful Life (RUL) of the vehicle components and including a predefined set of parameters for the vehicle; Based on the corresponding RUL of the vehicle component, assign the corresponding health status of the vehicle component to the dataset; Based on the changes that occur in response to changes in the health status of the vehicle components, the dataset is used to select a subset of parameters from the predefined parameter set; A machine learning model for detecting the health status of the vehicle components is trained using parameter values ​​from the subset of parameters in the plurality of datasets. The machine learning model is used to determine the health status of the vehicle component, wherein the health status of the vehicle component indicates the RUL of the vehicle component.

Images