Automotive physical layer (PHY) fault diagnosis

Abstract

To provide a method and a system for using an echo measurement value for detecting a cable failure in a cable of a vehicle communication network.SOLUTION: A method includes the steps of: obtaining at least one return echo signal corresponding to a signal communicated through a cable in a vehicle; applying, to the at least one return echo signal, a model trained by a computer for distinguishing a plurality of predetermined fault types in the cable from among at least two different predetermined fault types in the cable in order to identify the predetermined fault types; and initiating operation in response to the identified predetermined fault type, based on characteristics of the return echo signal learned from computer analysis of many previous echo signals.SELECTED DRAWING: Figure 3

Description

Technical Field

[0001] [Cross - Reference to Related Applications] This application claims the benefit of co - owned U.S. Provisional Application No. 63 / 118,484, filed on November 25, 2020, the disclosure of which is incorporated herein by reference.

[0002] The present disclosure generally relates to in - vehicle networks, and more particularly to methods and systems for diagnosing faults in the cables of in - vehicle networks.

Background Art

[0003] Modern vehicles, particularly autonomous vehicles, operate while generating a very large amount of data that is sensed or otherwise detected, analyzed, and transmitted through in - vehicle networks via cables. Faults in these cables can result in the in - vehicle network functioning at a reduced speed, malfunctioning, or not functioning at all. Faults in cables typically result from aging and physical damage. If a cable fails or becomes inappropriately functional, the vehicle may become unsafe or inoperable, and repairs to the vehicle and in - vehicle network are costly.

[0004] The above description is presented as a general overview of related technology in the art and should not be construed as an admission that any of the information included in the above description constitutes prior art against this patent application.

Summary of the Invention

[0005] The subject matter disclosed herein relates to a method and system for using echo measurements to detect cable faults in vehicle communication network cables. The system and method described above can detect a number of cable fault types, including short-circuits to ground and short-circuits to power supplies, which were previously undetectable due to system design constraints. In addition, the method and system can be easily integrated into existing vehicle communication networks, as no special hardware is required when using the method or system with existing vehicle communication networks.

[0006] Embodiments of the subject matter disclosed relate to a method for fault diagnosis in one or more cables of a vehicle communication network. The method is The steps include obtaining at least one return echo signal corresponding to a signal communicated through a cable in a vehicle, The steps include applying a computer-trained model for distinguishing between multiple predetermined fault types in a cable to at least one of the return echo signals in order to identify a predetermined fault type from among at least two different predetermined fault types in the cable, based on the characteristics of the return echo signal learned from computer analysis of a large number of previous echo signals, A step in which operation is initiated according to the identified predetermined fault type. Includes.

[0007] In some embodiments, a method for fault diagnosis is a method in which the step of applying a computer-trained model includes the step of applying an artificial intelligence (AI) model that has been pre-trained to distinguish the predetermined fault type in the cable from among the at least two predetermined fault types.

[0008] In some embodiments, the fault diagnosis method is a method in which the AI ​​model includes a plurality of feature extraction layers configured to determine that there is a fault in the cable and to identify the predetermined fault type. In some embodiments, the fault diagnosis method is a method in which the step of acquiring the at least one return echo signal is a step of receiving the at least one return echo signal by an Ethernet® physical layer (PHY) device, wherein the Ethernet® PHY device applies echo cancellation in response to the step of receiving the at least one return echo signal. In some embodiments, the fault diagnosis method is a method in which the characteristics of the return signal include the shape of the return echo signal, and the step of applying the computer-trained model includes a step of analyzing the shape of the return echo signal in comparison to predetermined shapes of at least two return echo signals, wherein each of the predetermined shapes is associated with a predetermined fault type from at least two different predetermined fault types.

[0009] In some embodiments, a fault diagnosis method is a method in which the step of applying the computer-trained model to the at least one return echo signal is determined by determining the probability that the one or more cables exhibit the identified predetermined cable fault type selected from at least two different predetermined cable fault types. In some embodiments, a fault diagnosis method is a method in which the AI ​​model includes one or more of the trained neural networks and / or machine learning models. In some embodiments, a fault diagnosis method is a method in which the plurality of predetermined fault types in the cable are Short circuit (short circuit), Open circuit (open), Open and grounded, Normal and grounded, One wire short-circuited to the ground. One wire short-circuited to the power supply. One wire is released, Two wires short-circuited to the ground. Two wires short-circuited to the power supply. and two wires are released This method involves two or more of the following:

[0010] In some embodiments, the fault diagnosis method is such that the at least one return echo signal comprises a plurality of return echo signals, each return echo signal is obtained from a corresponding cable, and the step of applying the computer-trained model comprises the step of analyzing each return echo signal to identify a predetermined fault type in each cable from at least two predetermined fault types. In some embodiments, the fault diagnosis method is such that the step of initiating the operation comprises one or more of the steps of logging the fault, issuing a warning, notifying a driver or other user, and switching to a backup cable.

[0011] Embodiments of the subject matter disclosed relate to a system for fault diagnosis in one or more cables of a vehicle communication network. The fault diagnosis system comprises one or more processors and a program memory for storing executable instructions. When an executable instruction is executed by the one or more processors, the system will In a vehicle, obtain at least one return echo signal corresponding to a signal communicated via a cable, Based on the characteristics of the return echo signal learned from computer analysis of numerous previous echo signals, in order to identify a predetermined fault type from among at least two different predetermined fault types in the cable, a computer-trained model for distinguishing between multiple predetermined fault types in the cable is applied to at least one return echo signal, To initiate operation according to the identified fault type This is an executable instruction that causes the following to happen.

[0012] In some embodiments, the fault diagnosis system is such that the executable instruction causes the system to further analyze the characteristics of the return echo signal, including the shape of the return echo signal, by applying the computer-trained model to determine the probability of the identified predetermined fault type from at least two different predetermined fault types in the cable. In some embodiments, the fault diagnosis system is such that the computer-trained model includes an artificial intelligence (AI) model stored in the program memory that determines the probability of the identified predetermined fault type from at least two different predetermined fault types in the cable. In some embodiments, the fault diagnosis system is such that the AI ​​model includes one or more of a trained neural network and / or machine learning models.

[0013] Embodiments of the subject matter disclosed relate to Ethernet® physical layer (PHY) devices for cable-based communication in automotive or industrial networks. The Ethernet® PHY device is The cable includes a transmitting unit for transmitting outbound Ethernet (registered trademark) communication signals, A receiving unit for receiving an inbound Ethernet® communication signal from the cable, wherein the inbound Ethernet® communication signal includes the return echo signal caused by the transmitted outbound Ethernet® communication signal, An echo canceller for estimating the return echo signal and for canceling the return echo signal in the received inbound Ethernet® communication signal, A storage medium communicating with the echo canceller for storing the estimated return echo signal, wherein the stored estimated return echo signal is available for reading by a device outside the Ethernet (registered trademark) PHY device. Comprises. In some embodiments, the Ethernet (registered trademark) PHY device is an Ethernet (registered trademark) PHY device such that the return echo signal is of a predetermined size.

[0014] This disclosure will be more fully understood from the following detailed description of its embodiments in light of the drawings.

Brief Description of the Drawings

[0015] [Figure 1A-1] FIG. 14 is a block diagram of an exemplary in-vehicle network (IVN) in a vehicle operating according to an embodiment described herein.

[0016] [Figure 1A-2] Insertion figure of FIG. 1A-1.

[0017] [Figure 1B] FIG. 26 is a diagram of a system outside the vehicle and an in-vehicle network (IVN) operating according to an embodiment described herein.

[0018] [Figure 2] FIG. 32 is a diagram of an exemplary physical layer (PHY) transceiver used to obtain return echo data measurements, particularly according to an embodiment described herein.

[0019] [Figure 3] FIG. 38 is a flowchart of an exemplary process according to an embodiment described herein.

[0020] [Figure 4]This is a block diagram schematically showing the system architecture for determining cable failures and failure types in the in-vehicle network shown in Figure 1A-1, according to one embodiment described herein.

[0021] [Figure 5A] This is a block diagram of an exemplary neural network structure for performing cable fault detection and identification according to one embodiment described herein.

[0022] [Figure 5B] Figure 5A shows an exemplary layer diagram of the neural network structure according to one embodiment described herein.

[0023] [Figure 6] This is a diagram showing cable failure types and their corresponding echoes.

[0024] [Figure 7] This figure shows the training speed for a neural network model according to one embodiment described herein.

[0025] [Figure 8A] This is a flowchart of the process performed by the system shown in Figure 4, according to one embodiment described herein. [Figure 8B] This is a flowchart of the process performed by the system shown in Figure 4, according to one embodiment described herein. [Modes for carrying out the invention]

[0026] Figure 1A-1 shows a vehicle 100, for example, an automobile, having an in-vehicle network (IVN) 101, also known as a vehicle communication network, and the terms vehicle communication network and in-vehicle network are used interchangeably herein. Vehicle 100 may be a standard non-autonomous vehicle or an autonomous vehicle, classified by one of the Society of Automotive Engineers (SAE) levels: 0 No automation, 1 Driver assistance, 2 Partial automation, 3 Conditional automation, 4 High automation, 5 Full automation. The vehicle communication network includes, for example, integrated circuit (IC) chips for performing various functions. There are also multiple cables, which connect various components of the vehicle communication network. With this arrangement of the IVN, the subject matter disclosed operates at the link level and the network level.

[0027] The vehicle communication network 101 includes a processor 110, for example, a central processing unit (CPU), a graphics processing unit (GPU), or other suitable computer processing unit, where the processor 110 represents an in-vehicle computer, also known herein as a local computer. The processor 110 communicates directly or indirectly with the controller 112 and the switch / gateway 114 via a physical layer device (PHY) 116 and an Ethernet® link 118a. The switches / gateway 114 communicate with each other directly or indirectly via the Ethernet® link 118a and the PHY 116. The switches / gateway 114 communicate with the PHY 116 directly or indirectly via an Ethernet® link 118b.

[0028] Links 118a and 118b include cables for data transmission and / or power transmission, for example, also known as Ethernet® cables. Links 118a and 118b are also hereafter referred to as "cables" or "Ethernet® cables." Cable types may be shielded or unshielded, single-core or multi-core, and multi-core cables may be, for example, twisted-pair cables. Data transmission capabilities may not be the same for all cable types. For example, in one embodiment, Ethernet® link 118a supports data transfer rates of 10 / 25 Gbps, while Ethernet® link 118b supports data transfer rates of 2.5 / 5 / 10 Gbps.

[0029] Each PHY 116 links a link partner, typically between an electronic unit and a local computer, i.e., a processor 110, via a suitable media access control (MAC) device. The electronic unit includes, for example, a camera 120, a radar / lidar / sonar 122, or other suitable sensors 124, such as a temperature sensor and a magnetic field sensor. The PHY 116 communicates with the local computer, i.e., a processor 110, via cables, typically through switches 114 and a controller 112.

[0030] Each link 118a, 118b has at least one end connected to switch 114. Typically, an echo signal acquired at one end of a link is sufficient to diagnose the Ethernet® cable. Thus, in one embodiment, switch 114 reads the echo signal from, for example, the memory 220 (Figure 2) of the PHY 116 to which switch 114 is locally connected, and relays the echo signal to processor 110 in the form of an echo measurement value representing the returned echo signal. A trained computer model 400, such as an artificial intelligence (AI) model, downloaded by and executed in processor 110, receives this echo measurement value as input.

[0031] In one embodiment, PHY116 receives echo signals, such as return echo signals, from cables 118a and 118b, estimates the corresponding return echo signals, and returns the estimated return echo signals via switch 114, for example, as echo measurements, through Ethernet® cables 118a and 118b. An exemplary PHY116 for receiving return echo signals and estimating echo cancellation signals as echo measurements is shown, for example, in Figure 2 and described in detail below.

[0032] As seen in insertion 130 of Figure 1A-2, for the processor 110, the processor 110 downloads a model 400, for example, a rule-based (non-AI) model or an AI model. The processor 110 is also associated with at least one transmit / receive unit 132, as described below, to receive a return echo signal and / or data associated with the return echo signal, input the signal and / or data to the model 400, and transmit the output generated by the model as a metric to other system computers, processors, and components, etc.

[0033] Figure 1B shows a system 140 located outside the vehicle 100 and IVN 101 for detecting cable faults in Ethernet® cables 118a and 118b connected to the PHY 116 of the vehicle 100. The system 140 includes a computer 142 which has downloaded an artificial intelligence (AI) model 400 or other downloadable model, including non-AI models (such as rule-based models), and receives echo responses from the PHY 116, such as echo cancellation signals and / or data associated with the echo cancellation signals, to determine whether there is a fault in the Ethernet® cables 118a and 118b, and if there is a fault, to determine the type of cable fault.

[0034] The configuration of the IVN101 shown in Figure 1A-1, and the composition and / or deployment of network components such as the processor 110, controller 112, switch 114, PHY 116, and various other system elements, such as the camera 120, radar / lidar / sonar 122, or other suitable sensors 124 such as temperature and magnetic field sensors, are illustrative configurations shown for illustrative purposes only. In alternative embodiments, any other suitable configuration may be used.

[0035] Various elements of the IVN101, such as the processor 110, controller 112, switch 114, PHY 116, and other system elements, such as the camera 120, radar / lidar / sonar 122, or other suitable sensors 124 such as temperature and magnetic field sensors, may be implemented using dedicated hardware or firmware in, for example, one or more application-specific integrated circuits (ASICs) and / or one or more field-programmable gate arrays (FPGAs), for example, by using hardwired logic or programmable logic. Additionally or alternatively, some functions of the above-mentioned components of the IVN may be implemented in software and / or using a combination of hardware and software elements. Elements not essential for understanding the disclosed technology have been omitted from the diagram for clarity.

[0036] In some embodiments, the processor 110 of an internal computer and / or the processor of an external computer 142 (which may be one or more computers) include one or more programmable processors programmed in software to perform the functions described herein. The software may be downloaded to any of these processors in electronic form, for example, over a network, or alternatively or additionally, provided and / or stored on a non-temporary tangible medium such as magnetic memory, optical memory, or electronic memory.

[0037] Figure 2 shows an exemplary PHY 116 represented as circuit 200 according to one embodiment described herein. The PHY 116 in Figure 2 performs echo cancellation for internal use (to improve receiving performance).

[0038] In this example, PHY116 includes a data modulator 202 that outputs an Ethernet® signal containing data symbols. Transmitter (TX) 204 transmits the Ethernet® signal to cable 118a / 118b via a hybrid coupling 206 (a hybrid coupling for splitting and combining signals to / from different channels). The transmitted data is directed to peer PHY116 (not shown) located further down the cable.

[0039] Upon reception, the Ethernet® signal from peer PHY116 is received via the hybrid coupling 206 through cables 118a and 118b and supplied to filter 208. Filter 208 filters the received signal, for example, using a low-pass filter or a band-pass filter, to remove undesirable frequency components. The analog-to-digital converter (ADC) 210 digitizes (samples) the signal. For example, the ADC 210 may digitize the received signal in batches of 16 phases, each having 48 samples per phase, for a total of 768 samples.

[0040] The digitized signal is filtered by a feedforward equalizer (FFE) 212. A decision circuit 214 (e.g., a slice circuit) decodes the bits from the equalized signal. The decoded bits are passed through a physical coding sublayer (PCS, not shown) and then provided as an output to, for example, a medium access control (MAC) device (not shown). For example, the MAC device may be switch 114. In this example, the signal is also equalized by a decision feedback equalizer (DFE) 216, whose input is taken from the output of the decision circuit 214.

[0041] PHY116 further includes an echo canceller (EC) 218, which cancels echoes of the transmitted Ethernet® signal that may leak into the received Ethernet® signal and thus interfere with the received Ethernet® signal. Typically, the EC 218 estimates the time waveform (shape in the time domain) of the echo signal based on the transmitted signal and cancels the echo signal from the received signal. Generally speaking, the EC 218 generates a replica of the transmitted Ethernet® signal with a magnitude and phase that matches the estimated echo signal, and then subtracts this replica from the received Ethernet® signal. The cancellation operation is illustrated by the adder 219 in the figure.

[0042] In one embodiment, the PHY 116 further comprises a storage medium, for example, a memory 220. The EC 218 stores the estimated echo signal as a vector of digital values ​​in the memory 220. The memory 220 is such that the echo signal is accessible for reading by any suitable external processor or system (e.g., by processor 110 or by computer 142) for use in detecting and classifying potential faults in cables 118a, 118b. In an exemplary embodiment, though not necessarily so, the estimated echo signal includes 768 digital values ​​sampled using 16 phases and a sampling rate of 66.66 MHz, resulting in a resolution of 0.9376 nS per sample.

[0043] As shown in the figure, FFE212, DFE216, and EC218 are controlled by suitable adaptation algorithms that run on a suitable processor of a PHY device (not shown). In this example, the input to these algorithms is an error signal (shown as err in the figure) indicating the difference between the received signal and the respective bit decision.

[0044] Either the local computer or the internal computer, i.e., the processor 110, or the computer 142 of the external cable fault detection system 140 (also referred to herein as “external computer 142”), performs exemplary steps of cable fault detection and identification of the detected cable fault type based on the echo signal reported by the PHY 116. An exemplary step is roughly shown in Figure 3, which is of interest here.

[0045] In Figure 3, the process begins in block 300. At this starting point, in one embodiment, the PHY116 or IVN101 is in diagnostic mode, ready to perform cable diagnostics. The process proceeds to block 302, where a model, e.g., a computer-trained model, is acquired to detect cable faults and identify the type of cable fault detected in the cable. The model 400 helps distinguish between several predetermined fault types (cable fault types) in the cable by identifying the predetermined fault types based on characteristics of the return echo signal, e.g., shape (and / or data associated with the return echo signal, known as "signal data"), and the predetermined fault types are learned from a computer analysis of a number of previous echo signals, e.g., from at least two different predetermined fault types in the cable.

[0046] For example, Model 400 may be a pre-trained neural network model. This is typically implemented by a local computer 100 or an external computer 142, which downloads the pre-trained neural network model from a location along the network and / or from the cloud. The process proceeds to block 304, where signal data corresponding to the return echo signals received by PHY 116 from one or more cables 118a, 118b is obtained from PHY 116 as needed. For example, PHY 116 also uses this signal data to form an echo cancellation signal used internally by PHY 116, for example, by estimation.

[0047] The process proceeds to block 306, where the acquired signal data is applied to the trained model. The trained model analyzes the signal data to determine whether there is a fault in the corresponding cable, and if a fault is detected, identifies the fault, including classifying it as a type of cable fault (cable fault type) from among known types of cable faults. In the optional process of block 306, the trained model may also determine the probability of the cable fault type from at least two predetermined fault types, or from at least two predetermined sets of fault types.

[0048] For cables analyzed by the model, the detected fault type, the predetermined faults identified for the detected fault type, and optionally, the probability of the detected fault type from at least two predetermined fault types are metrics output by the model in block 308.

[0049] The process proceeds to block 310, where, based on and in accordance with the outputted metrics, one or more actions are initiated and performed based on the identified fault type. For example, the actions initiated may be one or more of the following: logging the fault, issuing a warning, notifying the driver or other user associated with the vehicle 100, and switching to a backup cable.

[0050] The process then proceeds to block 312, where the aforementioned signal data is optionally uploaded based on detected cable faults and their identification information, in order to retrain or enhance the training of the trained model. The process ends in block 314 and may be repeated indefinitely as requested.

[0051] Figure 4 is an architectural block diagram of a trained model 400, such as a neural network model (such as an artificial intelligence (AI) model or a machine learning (ML) model), downloaded to a vehicle or local computer, i.e., typically one of the processor 110 or an external computer 142, according to an embodiment of the present disclosure. The trained model 400 (e.g., a computer-trained model) receives input, which is return echo signal data 410, such as data associated with the return echo signal from cables 118a, 118b, and the PHY 116 receives the return echo signal through Ethernet® cables 118a, 118b. The echo cancellation signal is expressed to the Ethernet® cable 118b of the IVN 101 as a value or data representing the signal shape. The echo cancellation signal and the corresponding signal data are typically generated at predetermined time intervals, typically when the vehicle 100 is started. The example described above is for signal shape in the time domain. Alternatively, the techniques described may also be adapted to detect signal shape anomalies in the frequency domain.

[0052] In an alternative embodiment, the above-described signal data is received, for example, by the onboard 110 of the vehicle 100, or by an external computer 142, in response to periodic queries, or, for example, automatically on a periodic basis.

[0053] In one embodiment, the trained model 400 (e.g., a computer-trained model) may be downloaded remotely via an external network such as the Internet, and portions of the trained model 400 may be stored in the cloud 420. There is also an associated module for runtime recording 422. The trained model 400, running on the processor 110 or computer 142, outputs a set of metrics in response to input information, i.e., signal data representing the return echo signals to the Ethernet® cables 118a, 118b received by the PHY 116.

[0054] The set of metrics includes, for example, the detection of cable failures 430, the identification of the type of cable failure of the detected cable failure 432, and the probability of the detected cable failure type from among two or more types of cable failures 434 (e.g., a set of cable failure types). The results of these metrics are analyzed, for example, by the processor 110 or an external computer 142 to initiate actions to be taken in the vehicle based on the identified cable failure type 436. The actions to be initiated include, for example, logging the cable failure (and the type of cable failure), issuing a warning, notifying the vehicle driver or other parties associated with the vehicle, and switching to a backup cable.

[0055] The trained model may be a rule-based model. A rule-based model works, for example, by analyzing the major peaks of a signal. As shown in Figure 6, in box 602, a positive major peak indicates an open fault, and a negative major minimum indicates a short fault, while the cable is normal unless there is a major peak.

[0056] Now, we will focus on Figure 5A, which is a diagram of the network architecture for an exemplary trained model 400, a neural network model according to one embodiment of the present disclosure. The neural network model 400 is a seven-layer neural network (e.g., a seven-layer convolved feature extraction unit 502) having a bidirectional network structure (an output layer with a customized loss function, which will be described in detail below) for fault detection 504a and fault type identification, including classification 504b. The mean squared error (MSE) of the cross-entropy for the detection unit 504a and the classification unit 504b, with the loss function summed, is mathematically expressed below.

[0057] The input, i.e., signal data, is fed into layer 1 of the neural network model 400, and the output is received from the detection unit 504a and the classification unit 504b. The number of layers, the type of layers including hidden layers, and / or the size of the layers depend, for example, on the number of transformations performed and the complexity of the transformations.

[0058] In one embodiment, each layer (Conv1D to Conv7D) in the feature extraction unit 502 is, for example, a complete convolutional layer 510, as shown by layer 1 (Conv1D) in Figure 5B. For example, as shown in Figure 5A, the size of the first layer (CONV1D_1) is 394 × 64, the size of the second layer (CONV1D_2) is 192 × 16, the size of the third layer (CONV1D_3) is 96 × 6, the size of the fourth layer (CONV1D_4) is 48 × 16, the size of the fifth layer (CONV1D_5) is 24 × 8, the size of the sixth layer (CONV1D_6) is 12 × 8, and the size of the seventh layer (CONV1D_7) is 6 × 64. The neural network model 400 includes node activation functions, such as sigmoid or normalized linear unit (ReLU), to perform the desired transformation.

[0059] This particular neural network architecture is shown herein merely as an example, and the principles of this disclosure may be alternatively implemented in other types of neural networks having more or fewer layers, and more or fewer inputs, outputs, and nodes within each layer. The visualized neural network model 400 is applicable to Ethernet® cable fault determination and Ethernet® cable fault identification, which include classifying determined cable faults and determining the probability of a fault type for a determined fault from two or more (e.g., one set) predetermined cable fault types. Typically, a processor 110, or a processor in an external computer 142, runs multiple models of this kind to process cable fault information regarding multiple Ethernet® cables associated with communication links and other components.

[0060] For example, a neural network model 400 downloaded from the cloud or another centralized processor useful for multiple vehicles to the vehicle processor 110 or an external computer 142 is trained for the classification unit 504b according to an objective function. Since training typically consumes a lot of computing resources and a lot of power, training may preferably be performed in the cloud. For separate outputs, for example, the objective function for the classification unit 504b is based on cross-entropy (CE), which is expressed as follows:

number

[0061] For example, if the fault type is "open", then y1 to y4 are, with respect to the classification unit 504b, [p o ,p s ,p n&g ,p o&g It could also be [0.9,0,0,0.1] which corresponds to ].

[0062] For continuous outputs such as fault determination and fault identification, which include fault classification and determining the probability of a fault type from two or more predetermined cable fault types, a loss function is determined for Ethernet® cables. The loss function is based on the mean squared error (MSE) for the detection unit 504a and the cross-entropy (CE) for the classification unit 504b. The final or customized loss function (e.g., total loss (TL)) is expressed as a weighted sum using weights w1 and w2: TL = w1 × MSE + w2 × CE.

[0063] Any suitable algorithm may be used to train the neural network model 400 based on this customized loss function. For example, the algorithm may be the Adam optimization algorithm, or random initial weights may be used. The learning speed for the customized loss function can be controlled by adjusting the learning rate during training of the neural network model.

[0064] The trained neural network model 400 is trained, for example, on a remote server along the network or in the cloud. The neural network model is trained by receiving inputs from a number of vehicles (IVNs) different from vehicle 100, which are signal shapes indicating normal cables or various cable fault types, ranging from detected cable faults to normal cables. This input is uploaded to the neural network model 400. Training the neural network model 400 using inputs from many vehicles is typically centralized, for example, so that the trained model benefits from a large amount of input. Therefore, the trained model is useful for a given vehicle, such as vehicle 100 (e.g., typically individual vehicles).

[0065] For example, Model 400 is trained on known or predetermined signal shapes for various types of cable faults obtained from numerous IVNs in numerous vehicles. The predetermined echo signal shapes described above for normal cables and various cable fault types are shown, for example, in Figure 6. For an Ethernet® cable length of approximately 6 meters: 1) Box 602 shows the return echo signal shapes for normal (fault-free cable), open fault, or short-circuit fault in an Ethernet® cable. 2) Box 604 shows the return echo signal shapes for open fault, or open fault and ground fault in an Ethernet® cable. 3) Box 606 shows the return echo signal shapes for normal (fault-free cable) and normal and ground fault in an Ethernet® cable. Model 400 analyzes the shape of the return echo signal from the signal data in comparison to the shape of the signal for each cable fault type to determine whether or not there is a cable fault. If there is a cable fault, it identifies the type of cable fault, or if not identified, classifies it. For detected cable faults, it determines the probability of the type of cable fault from at least two different types of cable faults based on the signal shape.

[0066] For example, the trained neural network model 400 is trained based on the following: This is based on signal data collected over a period of time from a component, namely the PHY 116 of IVN 101, in a set or multiple vehicles, different from vehicle 100 (e.g., a given vehicle), which is downloaded to a local computer (e.g., processor 110) or an external computer 142 of the vehicle. For example, the model determines whether a predetermined fault exists in a cable, distinguishes between multiple predetermined fault types in a cable within a vehicle network, identifies the predetermined fault type in the cable based on the characteristics of the return echo signal in the cable, for example, the predetermined fault type is learned from a computer analysis of a number of previous echo signals from at least two different predetermined fault types (or from a set of at least two different predetermined fault types), and / or, optionally, determines the probability of the identified predetermined fault type in the cable from at least two different predetermined fault types (or from a set of at least two different predetermined fault types). For example, predetermined fault types in a cable include short circuit, open circuit, open and ground, normal and ground, one wire short to ground, one wire short to power, one wire open, two wires short to ground, two wires short to power, and two wires open.

[0067] The collected signal data is periodically uploaded to a centralized processor, for example, to train the trained neural network model 300, where the trained neural network model 400 resides. The location is a remote network from the vehicle 100, such as the vehicle's local computer 110, an external computer 142, or the cloud.

[0068] The neural network model 400 is trained to generate a set of metrics based on collected signal data, for example, as shown and explained in Figure 4. The set of metrics indicates the detected faults in cable 430, the type of cable fault (identified type) for the detected cable faults 432, and the probability of the fault type from two or more predetermined fault types 434 for the detected cable faults, along with the actions initiated in response to the detected and identified cable fault type 436.

[0069] The trained neural network model is deployed as a software package for installation on the local vehicle's computer, e.g., processor 110, or on an external computer 142 (e.g., a processor), for example (which may be downloadable when the vehicle is parked and connected to an external wired or wireless network connection). The trained neural network model is typically associated with an application programming interface (API) for updating the trained neural network model 400.

[0070] Figure 7 shows the training setup for neural network model 400. The training speed is based on optimization algorithms such as the loss function, combined mean squared error and cross-entropy, and root mean square propagation (RMSProp). The figure shows the adjustable learning rate for model 400, such as the learning speed (LR) (y-axis) contrasted with the reference time (x-axis), which is a time base value. Initially, training is performed at a rapid speed and then slows down over time.

[0071] We will now focus on Figures 8A and 8B, which show flowcharts illustrating the details of a computer-implemented process according to an embodiment of the subject matter disclosed. The process described above, including subprocesses, is executed, for example, automatically and / or in real time.

[0072] The process begins in start block 802, where the local computer 110 of the vehicle's IVN, i.e., a local computer associated with the vehicle communication network, such as a given vehicle or external computer 142, is activated. The process proceeds to block 804, where a model, for example, a computer-trained model, is acquired.

[0073] A computer-trained model distinguishes between several predetermined fault types in a cable within a vehicle communication network, and identifies a predetermined fault type in the cable, learned from computer analysis of numerous previous echo signals, from at least two different predetermined fault types (or a set of at least two different predetermined fault types), based on the characteristics of the return echo signal in the cable, and / or, optionally, determines the probability of the identified predetermined fault type in the cable from at least two different predetermined fault types (or a set of at least two different predetermined fault types). For example, predetermined fault types in a cable include short circuits, open circuits, open and ground, normal and ground, one wire short to ground, one wire short to power, one wire open, two wires short to ground, two wires short to power, and two wires open.

[0074] The characteristics of the return echo signal include, for example, the shape of the return echo signal. The step of applying a computer-trained model includes, for example, the step of analyzing the shape of the return echo signal against at least two predetermined shapes of return echo signals, where each of the predetermined echo signal shapes is associated with a predetermined cable fault type from at least two different predetermined fault types.

[0075] Additionally, at least one acquired return echo signal may be, for example, multiple return echo signals, each of which is acquired from the corresponding cable of the IVN101. The step of applying a computer-trained model includes analyzing each return echo signal to identify a predetermined cable fault type in each cable from at least two predetermined fault types.

[0076] The computer-trained model may be, for example, an artificial intelligence (AI) model that is pre-trained to distinguish a predetermined type of failure in a cable from at least two predetermined failure types (or from at least two different sets of predetermined failure types). The AI ​​model may include, for example, multiple feature extraction layers, each configured to identify failure types so as to determine that there is a failure in the cable and that the detected failure is identified as one of the predetermined failure types. The AI ​​model may include, for example, one or more pre-trained neural networks and / or machine learning models.

[0077] The process proceeds to block 806, where the computer-trained model is downloaded to a local computer associated with the vehicle communication network, for example, local computer 110 or external computer 142 associated with the vehicle communication network of a given vehicle. The computer-trained model is then deployed in block 808 by the local computer associated with the vehicle communication network, for example, local computer 110 or external computer 142 of a given vehicle. The process proceeds to block 810, where at least one return echo signal (and / or data associated with at least one return echo signal) corresponding to a signal communicated through a cable in a given vehicle is acquired.

[0078] The process proceeds to block 812, where at least one return echo signal (and / or return echo signal data) from a given vehicle's cable is input into, or applied if no input is received, into, a computer-trained model. Next, in block 814, the output from the model is received, and the output is based on the actions performed by the computer-trained model, corresponding to at least one return echo signal (and / or return echo signal data) input to the computer-trained model for a given vehicle.

[0079] The process proceeds to block 816, where the response to the received output is made by initiating the following actions depending on the identified predetermined fault type, the actions including one or more of the following steps: logging the identified predetermined cable fault, issuing a warning, notifying the driver or other user associated with a given vehicle, and switching to a backup cable.

[0080] Moving on to block 818, return echo signals having the characteristics of known, predetermined cable fault types are provided (e.g., periodically) to update and / or add to at least two different fault types (e.g., at least two different predetermined sets of fault types). Then, in block 820, the characteristics of the return echo signals from one or more of a given vehicle and / or other vehicles are provided (e.g., periodically) to a computer-trained model for further training of the model.

[0081] The process proceeds to block 822 and terminates. The process may be repeated indefinitely as required.

[0082] The processor 110 (of the vehicle's local computer) or the external computer 142 may include a general-purpose computer programmed in software, for example, a trained neural network model, to perform the functions described herein. The software may be downloaded to the computer in electronic form, for example, over a network, or alternatively or additionally, provided and / or stored on a non-temporary tangible medium such as magnetic memory, optical memory, or electronic memory.

[0083] While the embodiments described herein primarily concern applications to automobiles (e.g., cars or trucks), the methods and systems described herein may also be used in other types of vehicle networks, such as in aircraft and boats.

[0084] It should be noted that the embodiments described above are cited as examples, and that the present invention is not limited to those specifically shown and described above. Rather, the scope of the present invention includes both combinations and subcombinations of the various features described above, as well as variations and modifications thereof that are not disclosed in the prior art and that a person skilled in the art would conceive of by reading the above description. Documents incorporated into this patent application by reference shall be considered integral parts of this application. However, insofar as any term is defined in such incorporated documents in a manner that is inconsistent with any definition made expressly or implicitly herein, only the definitions herein shall be considered.

Claims

1. A method for fault diagnosis in one or more cables of a vehicle communication network, wherein the method is The steps include obtaining at least one return echo signal corresponding to a signal communicated through a cable in a vehicle, The steps include applying a computer-trained model for distinguishing between multiple predetermined fault types in a cable to at least one of the return echo signals in order to identify a predetermined fault type from among at least two different predetermined fault types in the cable, based on the characteristics of the return echo signal learned from computer analysis of a large number of previous echo signals, A step in which operation is initiated according to the identified predetermined fault type. Includes, The characteristics of the return signal include the shape of the return echo signal, The step of applying the computer-trained model includes a step of analyzing the shape of the return echo signal in comparison with at least two predetermined shapes of the return echo signal, wherein each of the predetermined shapes is associated with a predetermined fault type from at least two different predetermined fault types. method.

2. The method for fault diagnosis according to claim 1, wherein the step of applying the computer-trained model comprises applying an artificial intelligence (AI) model that has been pre-trained to distinguish the predetermined fault type in the cable from among the at least two predetermined fault types.

3. The method for fault diagnosis according to claim 2, wherein the AI ​​model includes a plurality of feature extraction layers configured to determine that there is a fault in the cable and to identify a predetermined type of fault.

4. A method for fault diagnosis according to any one of claims 1 to 3, wherein the step of acquiring the at least one return echo signal is a step of receiving the at least one return echo signal by an Ethernet® physical layer (PHY) device, and the Ethernet® PHY device applies echo cancellation in accordance with the step of receiving the at least one return echo signal.

5. A method for fault diagnosis according to any one of claims 1 to 4, wherein the step of applying the computer-trained model to the at least one return echo signal is a step of determining the probability that the one or more cables exhibit the identified predetermined cable fault type selected from at least two different predetermined cable fault types.

6. The method for fault diagnosis according to claim 2 or 3, wherein the AI ​​model includes one or more of a trained neural network and / or a machine learning model.

7. The aforementioned multiple predetermined fault types in the cable are One wire short-circuited to the ground. One wire short-circuited to the power supply. One wire was released. Two wires short-circuited to the ground. Two wires short-circuited to the power supply, and The two wires are released. A method for fault diagnosis according to any one of claims 1 to 6, comprising two or more of the above.

8. The method for fault diagnosis according to any one of claims 1 to 7, wherein the at least one return echo signal comprises a plurality of return echo signals, each return echo signal is obtained from a corresponding cable, and the step of applying the computer-trained model comprises the step of analyzing each return echo signal to identify a predetermined fault type in each cable from among at least two predetermined fault types.

9. The method for fault diagnosis according to any one of claims 1 to 8, wherein the step of initiating the operation includes one or more of the steps of logging a fault, issuing a warning, notifying a driver or other user, and switching to a backup cable.

10. A system for diagnosing faults in one or more cables of a vehicle communication network, wherein the system One or more processors, When executed by the one or more processors, the system In a vehicle, obtain at least one return echo signal corresponding to a signal communicated via a cable, Based on the characteristics of the return echo signal learned from computer analysis of numerous previous echo signals, a computer-trained model for distinguishing between multiple predetermined fault types in a cable is applied to at least one return echo signal in order to identify a predetermined fault type from among at least two different predetermined fault types in the cable. In order to determine the probability of the identified predetermined failure type from among at least two different predetermined failure types in the cable, the computer-trained model is applied to analyze the characteristics of the return echo signal, including the shape of the return echo signal. To initiate operation according to the identified fault type Program memory that stores executable instructions and A system equipped with these features.

11. A system for diagnosing a fault according to claim 10, comprising an artificial intelligence (AI) model in which a computer-trained model is stored in the program memory and determines the probability of an identified predetermined fault type from among at least two different predetermined fault types in the cable.

12. The system for diagnosing a failure according to claim 11, wherein the AI ​​model includes one or more of the trained neural networks and / or machine learning models.

13. An Ethernet® physical layer (PHY) device for cable-based communication in an automotive or industrial network, wherein the Ethernet® PHY device is A transmitting unit for transmitting outbound Ethernet (registered trademark) communication signals to the aforementioned cable, A receiving unit for receiving an inbound Ethernet® communication signal from the cable, wherein the inbound Ethernet® communication signal includes a return echo signal caused by the transmitted outbound Ethernet® communication signal, An echo canceller for estimating the return echo signal and for canceling the return echo signal in the received inbound Ethernet (registered trademark) communication signal, A storage medium communicating with the echo canceller to store the estimated return echo signal, wherein the stored estimated return echo signal is available for reading by a device outside the Ethernet® PHY device. A processor configured to determine the probability of a predetermined failure type identified from at least two different predetermined failure types in the cable, based on the shape of the return echo signal, An Ethernet® PHY device equipped with [feature name].

14. The Ethernet® PHY device according to claim 13, wherein the return echo signal is of a predetermined size.

Images