A differential performance boundary test method, device, storage medium and vehicle

Abstract

This application discloses a method, apparatus, storage medium, and vehicle for differential performance boundary testing, relating to the field of differential performance testing technology. The method includes: extracting failure precursor factors from multimodal sensing data during the differential test in real time; determining the failure evolution path of the differential based on the multimodal sensing data in response to the failure precursor factors exceeding a preset failure threshold; determining an active intervention strategy corresponding to the failure evolution path; executing the active intervention strategy to determine the differential performance boundary test results. This application can capture abnormal signals in the differential failure initiation stage, creating a sufficient time window for active intervention, avoiding sudden damage to the prototype during testing, and simultaneously acquiring complete performance boundary data under extreme operating conditions to determine the differential performance boundary test results, thereby reducing R&D costs and shortening the testing cycle.

Description

Technical Field

[0001] This application relates to the field of differential performance testing technology, and in particular to differential performance boundary test methods, differential performance boundary test devices, storage media, and vehicles. Background Technology

[0002] The differential is one of the core components of a car's transmission system. Its main function is to allow the left and right drive wheels to rotate at different speeds when the vehicle is turning or when road conditions change, thereby ensuring the vehicle's stability and handling. Under extreme conditions, such as when one wheel slips on icy or snowy roads, when the vehicle is cornering at high speed, or when attempting off-road extrication, the speed difference between the planetary gears and half-shaft gears inside the differential increases dramatically, while simultaneously bearing complex alternating loads. If the differential fails under these conditions, such as gear seizure, burning, or even breakage, it will directly lead to a loss of vehicle power, and in severe cases, may cause a safety accident.

[0003] Currently, the durability verification of differentials mainly relies on bench tests of physical prototypes, which is commonly referred to in the industry as "differential limit differential operating condition baseline test". This type of test simulates extreme operating conditions on a test bench, gradually increasing the speed difference between the left and right output ends until the differential fails, thereby determining its limit performance boundary.

[0004] However, the above tests have limitations in understanding the failure modes of differentials. The detection methods are limited and have a lag, and they cannot predict the possible development trend before failure occurs. This often results in the test ending with "prototype damage", which not only wastes resources but also prolongs the research and development cycle. Summary of the Invention

[0005] The purpose of this invention is to provide a differential performance boundary test method, a differential performance boundary test device, an electronic device, a storage medium, and a vehicle, at least to solve the problem of how to overcome the limitations of monitoring a single physical quantity, and to solve a technical problem of how to predict the possible development trend of differential failure before it occurs.

[0006] This invention provides the following solution:

[0007] According to one aspect of the present invention, a method for testing the performance boundary of a differential is provided, comprising:

[0008] Extract failure precursor factors from multimodal sensing data during real-time differential testing;

[0009] In response to the failure precursor factor exceeding a preset failure threshold, the failure evolution path of the differential is determined based on the multimodal sensing data;

[0010] Determine the active intervention strategy corresponding to the failure evolution path;

[0011] The active intervention strategy is executed to determine the results of the differential performance boundary test.

[0012] Preferably, the extraction of failure precursor factors from the multimodal sensing data during the real-time differential test includes:

[0013] Data from multiple types of sensors deployed during the differential bench test are collected in real time to obtain multimodal sensing data of the differential bench test;

[0014] The encoder and decoder structures of the pre-trained autoencoder network are used to extract multimodal features from the multimodal sensing data.

[0015] Based on the multimodal features, the error of the differential is reconstructed to generate a normalized failure precursor factor.

[0016] Preferably, before determining the failure evolution path of the differential based on the multimodal sensing data, the method further includes:

[0017] In response to the failure precursor factor exceeding a preset abnormal threshold, the frequency of collecting the multimodal sensing data is increased.

[0018] Preferably, determining the failure evolution path of the differential based on the multimodal sensing data includes:

[0019] The differential equations describing physical laws in the differential performance boundary test process are incorporated into the physical information neural network to determine the physical information neural network failure evolution inference model.

[0020] Based on the physical information neural network failure evolution inference model, the multimodal sensing data is inferred to determine at least one failure evolution path;

[0021] Each failure evolution path corresponds to a failure probability time series.

[0022] Preferably, determining the active intervention strategy corresponding to the failure evolution path includes:

[0023] For each failure evolution path, the failure evolution path and its corresponding failure probability time series are determined as a set of failure evolution path data, resulting in multiple sets of failure evolution path data.

[0024] Multiple sets of failure evolution path data are input into a deep reinforcement learning decision model to generate the optimal active intervention strategy with the goal of minimizing failure risk and maximizing the value of experimental data.

[0025] Preferably, the step of generating the optimal proactive intervention strategy with the goal of minimizing failure risk and maximizing the value of experimental data includes:

[0026] With the goal of minimizing failure risk and maximizing the value of experimental data, the expected return of each set of failure evolution path data is evaluated on a millisecond timescale.

[0027] Determine the optimal expected return, and in at least one of the failure evolution paths, determine the first failure evolution path corresponding to the optimal expected return;

[0028] Determine the active intervention strategy corresponding to the first failure evolution path.

[0029] Preferably, the step of executing the active intervention strategy to determine the differential performance boundary test results includes:

[0030] The test bench controls the differential to execute the corresponding actions of the active intervention strategy to intervene in the failure of the differential until the failure precursor factor is less than or equal to the failure critical threshold.

[0031] The differential is then tested using the test parameters following the implementation of the active intervention strategy to determine the performance boundary test results of the differential.

[0032] According to a second aspect of the present invention, a differential performance boundary test apparatus is provided, comprising:

[0033] The failure factor extraction module is used to extract failure precursor factors in real time from the multimodal sensing data during the differential test process;

[0034] The failure evolution path determination module is used to determine the failure evolution path of the differential based on the multimodal sensing data in response to the failure precursor factor exceeding a preset failure critical threshold.

[0035] An intervention strategy determination module is used to determine the active intervention strategy corresponding to the failure evolution path;

[0036] The test result determination module is used to execute the active intervention strategy and determine the performance boundary test results of the differential.

[0037] According to three aspects of the present invention, an electronic device is provided, comprising: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0038] The memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the differential performance boundary test method.

[0039] According to four aspects of the present invention, a computer-readable storage medium is provided, comprising: storing a computer program executable by an electronic device, wherein when the computer program is run on the electronic device, the electronic device performs the steps of a differential performance boundary test method.

[0040] According to five aspects of the present invention, a vehicle is provided, comprising:

[0041] Electronic equipment used to implement the steps of a differential performance boundary test method;

[0042] The processor runs a program, and when the program runs, it executes the steps of the differential performance boundary test method based on the data output from the electronic device.

[0043] Storage medium for storing programs that, when run, execute steps of a differential performance boundary test method based on data output from electronic devices.

[0044] The above solution achieves the following beneficial technical effects:

[0045] This application extracts failure precursor factors from multimodal sensing data during real-time differential testing, enabling the capture of abnormal signals in the early stages of differential failure. This extends the warning time from seconds to minutes, creating a sufficient window for proactive intervention. Actual test data shows that detecting the differential's condition using failure precursor factors can provide effective warnings more than 45 seconds in advance, whereas traditional methods often only detect failures after they have occurred.

[0046] This application uses multimodal sensing data to determine the failure evolution path of the differential, enabling a visual simulation of the entire process of differential failure from initiation to outbreak. It allows for an intuitive view of the changing trends in failure probability under different intervention strategies, providing a quantitative tool for understanding failure mechanisms and optimizing product design.

[0047] This application avoids sudden damage to the prototype during testing by implementing an active intervention strategy, while also being able to obtain complete performance boundary data under extreme operating conditions, determine the performance boundary test results of the differential, reduce R&D costs, and shorten the testing cycle. Attached Figure Description

[0048] Figure 1 This is a flowchart of a differential performance boundary test method provided by one or more embodiments of the present invention.

[0049] Figure 2 This is a structural diagram of a differential performance boundary test device provided in one or more embodiments of the present invention.

[0050] Figure 3This is a block diagram of an electronic device for a differential performance boundary test method provided in one or more embodiments of the present invention. Detailed Implementation

[0051] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0052] In related technologies, extreme operating conditions are simulated on a test bench, and the speed difference between the left and right output ends is gradually increased until the differential fails, thereby determining its ultimate performance boundary. However, this process often only focuses on the final failure result, i.e., whether the differential is "damaged" and "the form of damage." However, there is a lack of effective monitoring methods and quantitative characterization capabilities for the complete process of failure from its inception to its final outbreak, including key intermediate stages such as how micro-damage accumulates, how the temperature field evolves, and how lubrication deteriorates. This prevents engineers from gaining a deep understanding of the failure mechanism and makes it difficult to guide targeted product optimization. Monitoring during the test mainly relies on a few physical quantities, such as oil temperature, input torque, or output shaft speed. The common practice is to set an alarm threshold, such as "alarm and shutdown when oil temperature exceeds 150°C." This single-threshold judgment method has significant shortcomings: firstly, the rise in oil temperature itself has a lag; by the time the oil temperature reaches the alarm value, the internal critical friction pairs may already be in a state of severe wear; secondly, a single physical quantity cannot reflect the complex characteristics of failure precursors under the coupling effect of multiple physical fields, easily leading to missed or false alarms. Extreme differential speed conditions represent a typical highly nonlinear, multivariate coupled process. Its failure development path is influenced by various factors, including temperature, load, lubrication condition, and wear accumulation. Current technologies cannot predict potential development trends before failure occurs, meaning they cannot answer critical questions such as "How long will it take for failure to occur if loading continues?" and "What intervention measures can effectively prevent failure?" This often results in tests ending in "prototype damage," wasting resources and prolonging the development cycle.

[0053] Based on this, this embodiment provides a differential performance boundary test method, device, storage medium, and vehicle to effectively monitor and quantify the entire differential failure process, achieving early identification of failure precursor characteristics under multi-physics coupling. It predicts the possible development trend before failure occurs and generates proactive intervention strategies accordingly, achieving a technological leap from "passive alarm" to "active control."

[0054] Figure 1This is a flowchart of a differential performance boundary test method provided by one or more embodiments of the present invention.

[0055] like Figure 1 The differential performance boundary test methods shown include:

[0056] Step S1: Extract failure precursor factors from the multimodal sensing data during the differential test in real time.

[0057] The multimodal sensing data during the differential test includes high-frequency vibration signals, acoustic emission signals, transient temperature gradient signals, and online physicochemical indicators of lubricating oil, forming a set of sensing data covering multiple physical fields such as heat, force, sound, and lubrication, thereby determining the multimodal sensing data.

[0058] The failure precursor factor is a dimensionless normalized index with a value range of [0, 1], used to quantify the early onset of a differential's evolution from normal operation to failure. Its calculation formula is as follows:

[0059]

[0060] In the formula, Indicates the precursor factors of failure. This is represented by the Sigmoid activation function, which maps the output to the [0,1] interval. The weight coefficients representing the i-th modal feature are obtained through training a deep autoencoder. The normalized eigenvalues ​​of the i-th mode include acoustic emission characteristics, temperature field characteristics, lubrication state characteristics, and vibration characteristics. This indicates the bias term.

[0061] Generally, the failure precursor factor range includes four values, corresponding to the four stages of the differential's evolution from normal operation to failure. The first range corresponds to the first stage of the differential, the normal operation stage, with a value less than 0.2, indicating that the differential is currently in a stable wear period without abnormal signs. The second range corresponds to the second stage, the early failure initiation stage, with a value greater than or equal to 0.2 and less than 0.5, indicating that the differential is currently in the failure incubation period, and microscopic damage has begun to accumulate. The third range corresponds to the third stage, the critical failure precursor stage, with a value greater than or equal to 0.5 and less than 0.8, indicating that the differential is currently in the failure initiation stage, and clear precursor characteristics can be detected. The fourth range corresponds to the fourth stage, the failure approach stage, with a value greater than or equal to 0.8. The typical failure threshold is 0.5, indicating that the differential is currently in the failure propagation stage and macroscopic failure is imminent.

[0062] Step S2: In response to the failure precursor factor exceeding the preset failure critical threshold, the failure evolution path of the differential is determined based on multimodal sensing data.

[0063] Based on the value range of the aforementioned failure precursor factors, the range of the currently extracted failure precursor factors is determined, thereby identifying the current stage of the differential.

[0064] If the currently extracted failure precursor factors exceed the preset failure threshold, the differential is determined to be in the critical failure precursor stage. The system immediately issues a secondary warning and determines the failure evolution path of the differential based on the currently collected multimodal sensing data. Furthermore, the failure evolution path of the current differential can be determined based on a pre-trained Physical Information Neural Network (PINN) failure evolution inference model.

[0065] Step S3: Determine the active intervention strategy corresponding to the failure evolution path.

[0066] In this embodiment, the physical information neural network failure evolution inference model can also generate a failure probability time series corresponding to the failure evolution path, thereby determining the active intervention strategy corresponding to the failure evolution path based on the failure evolution path and its corresponding failure probability time series.

[0067] Step S4: Implement the active intervention strategy to determine the results of the differential performance boundary test.

[0068] By implementing a defined active intervention strategy to address differential failure, the risk of differential failure can be mitigated within a certain timeframe, allowing for the completion of differential testing and determination of its energy boundary test results. Specifically, the automatic control system executes corresponding actions on the test bench and feeds the results back to the sensing module, forming a closed-loop control system.

[0069] This embodiment enables the detection of abnormal signals at the initial stage of failure, improving the warning time from the "second level" of traditional methods to the "minute level," creating a sufficient time window for proactive intervention. By introducing a physical information neural network (PINN), this invention achieves a visualized simulation of the entire process of differential failure from its inception to its outbreak, thereby preventing sudden damage to the prototype during testing through proactive intervention, while also enabling the complete acquisition of performance boundary data under extreme operating conditions.

[0070] In this embodiment, data from various sensors deployed during the differential bench test can be collected in real time to obtain multimodal sensing data from the differential bench test. Furthermore, multimodal features of the multimodal sensing data are extracted using the encoder and decoder structure of a pre-trained autoencoder network. These multimodal features include acoustic emission features, temperature field features, lubrication state features, and vibration features.

[0071] This embodiment integrates multi-dimensional features such as temperature field, vibration, acoustic emission, and lubrication state, and employs a deep autoencoder network for feature fusion, effectively overcoming the limitations of single threshold judgment. In verification experiments, the accuracy rate of this invention in identifying critical failure states reached over 92%, while the accuracy rate of related technologies was less than 60%.

[0072] The method for calculating acoustic emission characteristics is as follows:

[0073]

[0074] In the formula, Indicates acoustic emission characteristics, This indicates the energy percentage of the 280kHz characteristic frequency band. denoted as the statistical mean under normal operating conditions, and k represents the acoustic emission characteristic coefficient.

[0075] The calculation method for temperature field characteristics is as follows:

[0076]

[0077] In the formula, Indicates the characteristics of the temperature field. Indicates the rate of temperature rise. This indicates the temperature difference between local hotspots and the environment. Indicates the temperature coefficient. This represents the rate of temperature rise coefficient.

[0078] The calculation method for lubrication condition characteristics is as follows:

[0079]

[0080] In the formula, Indicates the characteristics of the lubrication state. This represents the particulate pollution weighting coefficient. This indicates the count of tiny particles in the oil, ranging from 2 to 5 μm. This indicates that the viscosity deviates from the weighting factor. In fact, viscosity sensitivity means This indicates the initial kinematic viscosity of the oil. This indicates the current kinematic viscosity of the oil. Indicates the sensitivity coefficient to the influence of particles. This represents the threshold for counting tiny particles.

[0081] The method for calculating vibration characteristics is as follows:

[0082]

[0083] In the formula, Indicates vibration characteristics, Indicates the effective value of high-frequency vibration. This represents the initial effective value of high-frequency vibration. This represents the vibration sensitivity coefficient.

[0084] Therefore, the differential error is reconstructed based on multimodal characteristics, and a normalized failure precursor factor is generated.

[0085] It should be noted that the extracted failure precursor factors are in the second range, that is, they exceed the preset abnormal threshold, but have not reached the failure critical threshold, thus increasing the frequency of collecting multimodal sensing data.

[0086] In this embodiment, the failure evolution path is implemented as follows:

[0087] The differential equations describing the physical laws during the differential performance boundary test are incorporated into the loss function of a physical information neural network to determine the failure evolution prediction model of the physical information neural network. The differential equations describing the physical laws include differential equations with physical properties such as heat conduction equations, the Arcard wear model, and elastohydrodynamic lubrication equations. The failure evolution prediction model of the physical information neural network is predetermined.

[0088] Based on the currently acquired multimodal sensing data and the determined physical information neural network failure evolution inference model, at least one failure evolution path is determined, and the failure probability time series corresponding to each failure evolution path is determined.

[0089] For each failure evolution path, the failure evolution path and its corresponding failure probability time series are defined as a set of failure evolution path data, resulting in multiple sets of failure evolution path data. These multiple sets of failure evolution path data are input into a deep reinforcement learning decision model to generate an optimal active intervention strategy with the goal of minimizing failure risk and maximizing the value of experimental data. The active intervention strategies include speed difference gradient adjustment, load spectrum adjustment, cooling flow rate adjustment, and emergency shutdown.

[0090] Furthermore, with the goal of minimizing failure risk and maximizing the value of experimental data, the expected return of each set of failure evolution path data is evaluated on a millisecond timescale. The optimal expected return is determined, and in at least one failure evolution path, the first failure evolution path corresponding to the optimal expected return is identified, along with the corresponding active intervention strategy.

[0091] For example, the test subject is a certain vehicle model with a maximum differential speed of 800 rpm. The test objective is to obtain the actual failure boundary of the differential under extreme differential conditions, so as to provide data support for the design optimization of the mass production version.

[0092] Multiple types of sensors were deployed on the differential test bench, including: three miniature thin-film thermocouples (0.5 mm in diameter, response time <5 ms) embedded in the planetary gear journal; an infrared thermal imager (640×512 resolution, 50 Hz frame rate) arranged in the differential housing window; an acoustic emission sensor (2 MHz sampling rate, 100 kHz-800 kHz bandwidth) installed; an online oil particle sensor (1 μm detection accuracy) connected; and a high-frequency acceleration sensor (±50 g range, 1 Hz-10 kHz frequency response) installed at the bearing housing location.

[0093] The test execution steps include: fixing the left output shaft and increasing the speed of the right output shaft from 0 to 800 rpm in a gradient of 10 rpm / s; holding each speed plateau (50 rpm interval) for 30 seconds and collecting steady-state data; if the system issues an early warning, adjusting the loading strategy according to the decision results.

[0094] The experimental procedure for executing the steps is as follows:

[0095] Phase 1: Normal operation (0-300rpm)

[0096] After the test began, all monitoring parameters were within the normal range. The oil temperature gradually increased from room temperature to 85℃, the vibration level remained stable, the acoustic emission signal was mainly background noise, and the oil particle count showed no significant change. The precursor factor remained below 0.08, and the system was in routine monitoring mode.

[0097] Phase Two: Early Abnormalities Appear (300-500 rpm)

[0098] When the speed difference reaches 300 rpm, a weak energy accumulation begins to appear in the acoustic emission signal at the 280 kHz frequency band. Simultaneously, the online oil sensor detects a slow increase in the number of 2-5 μm microparticles. At this point, traditional monitoring indicators (oil temperature, torque) still show normal readings, with the oil temperature at 105℃, far below the 150℃ alarm threshold. When the speed difference reaches 500 rpm, the energy percentage of the acoustic emission characteristic frequency band increases from 5% to 12% of the normal value, and the microparticle count increases from 3000 / mL to 8000 / mL. The precursor factor increases from 0.08 to 0.42, and the system determines that it has entered the "early germination" state, beginning to encrypt the data acquisition frequency.

[0099] Phase 3: Critical precursor triggering (500-600 rpm)

[0100] When the rotational speed difference continued to rise to 600 rpm, the temperature field reconstructed by the fusion of infrared thermal imaging and embedded thermocouples showed local hot spots at the meshing point of the planetary gear and the half-shaft gear, with a temperature difference of 52℃ from the environment and a temperature rise rate of 3.2℃ / s. The energy proportion of the acoustic emission 280kHz frequency band further increased to 18%, and the microparticle count exceeded 12,500 / mL. The normalized values ​​of each modal characteristic were: fAE=0.79, fTemp=0.85, fLube=0.58, fVib=0.87. After deep autoencoder fusion calculation, the precursor factor jumped to 0.64, exceeding the threshold of 0.5, and the system immediately triggered a level-two warning, activating the PINN failure evolution simulation module.

[0101] Phase Four: Evolutionary Deduction and Decision Making (650 rpm)

[0102] The PINN module is input with the following current state parameters: speed difference 650 rpm, local hotspot temperature 238℃, wear accumulation depth 0.02 mm, and temperature rise rate 3.2℃ / s. Based on the embedded physical equations, three possible evolution paths are deduced:

[0103]

[0104] The deep reinforcement learning agent completes decision calculations within 0.5 seconds. The current state has a precursor factor of 0.68, predicting a 78% probability of glue-gluing after 45 seconds. The agent evaluates the expected rewards of three actions:

[0105] Action 1 (Emergency Stop): The failure probability is reduced to 5%, but the test data is significantly lost, and performance data at higher speeds cannot be obtained;

[0106] Action 2 (reducing speed to 300 rpm): The failure probability is reduced to 28%, but excessively large gradient loading may cause data discontinuity;

[0107] Action 3 (cooling flow +40%, loading gradient reduced to 5 rpm / s): failure probability reduced to 15%, data integrity is good.

[0108] The agent selects action 3 and executes the instruction: increase cooling flow by 40% and reduce the speed difference loading gradient from 10 rpm / s to 5 rpm / s.

[0109] Phase 5: Risk Relief and Test Completion (650-800 rpm)

[0110] After the intervention was implemented, the local hotspot temperature dropped from 238°C to 205°C within 20 seconds, and the precursor factor fell back to 0.31. The system lifted the warning and continued the experiment with the optimized loading strategy.

[0111] Ultimately, the test successfully completed the entire loading process from 650 rpm to 800 rpm, and obtained complete performance boundary data of the differential under extreme differential conditions, including key information such as temperature field distribution, vibration characteristics, and lubrication state evolution at each speed difference.

[0112] To verify the technical effectiveness of this invention, a comparative experiment was conducted on the same test bench using conventional methods. The comparison results are as follows:

[0113]

[0114] Based on the test results of the above comparative experiment, it can be seen that this embodiment can overcome the limitations of traditional single physical quantity monitoring and realize early identification of the failure initiation stage.

[0115] Figure 2 This is a structural diagram of a differential performance boundary test device provided in one or more embodiments of the present invention.

[0116] like Figure 2 The differential performance boundary test device shown includes: a failure factor extraction module, a failure evolution path determination module, an intervention strategy determination module, and a test result determination module.

[0117] The failure factor extraction module is used to extract failure precursor factors in real time from the multimodal sensing data during the differential test process;

[0118] The failure evolution path determination module is used to determine the failure evolution path of the differential based on the multimodal sensing data in response to the failure precursor factor exceeding a preset failure critical threshold.

[0119] An intervention strategy determination module is used to determine the active intervention strategy corresponding to the failure evolution path;

[0120] The test result determination module is used to execute the active intervention strategy and determine the performance boundary test results of the differential.

[0121] The failure factor extraction module is used to collect data from multiple types of sensors deployed during the differential bench test in real time to obtain multimodal sensing data of the differential bench test; extract multimodal features of the multimodal sensing data through the encoder and decoder structure of the pre-trained autoencoder network; reconstruct the error of the differential based on the multimodal features to generate normalized failure precursor factors.

[0122] Specifically, in response to the failure precursor factor exceeding a preset abnormal threshold but not exceeding the failure critical threshold, the frequency of collecting the multimodal sensing data is increased.

[0123] The failure evolution path determination module is used to incorporate the differential equations describing physical laws in the differential performance boundary test process into a physical information neural network to determine a physical information neural network failure evolution inference model; based on the physical information neural network failure evolution inference model, the module infers the multimodal sensing data to determine at least one failure evolution path; wherein each failure evolution path corresponds to a failure probability time series.

[0124] The failure evolution path determination module is used to determine the failure evolution path and its corresponding failure probability time series as a set of failure evolution path data for each failure evolution path, thereby obtaining multiple sets of failure evolution path data; inputting the multiple sets of failure evolution path data into a deep reinforcement learning decision model, with the goal of minimizing failure risk and maximizing the value of experimental data, to generate the optimal active intervention strategy.

[0125] The intervention strategy determination module is used to evaluate the expected return of each set of failure evolution path data on a millisecond timescale with the goal of minimizing failure risk and maximizing the value of experimental data; determine the optimal expected return; and in at least one failure evolution path, determine the first failure evolution path corresponding to the optimal expected return; and determine the active intervention strategy corresponding to the first failure evolution path.

[0126] The test result determination module is used to control the test bench of the differential, execute the corresponding actions of the active intervention strategy, intervene in the failure of the differential until the failure precursor factor is less than or equal to the failure critical threshold; continue to test the differential with the test parameters after executing the active intervention strategy, and determine the performance boundary test results of the differential.

[0127] Figure 3This is a block diagram of an electronic device for a differential performance boundary test method provided in one or more embodiments of the present invention.

[0128] like Figure 3 As shown, this application provides an electronic device, including: a processor, a communication interface, a memory, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus;

[0129] The memory stores a computer program that, when executed by a processor, causes the processor to perform the steps of a differential performance boundary test method.

[0130] This application also provides a computer-readable storage medium storing a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of a differential performance boundary test method.

[0131] This application also provides a vehicle, including:

[0132] Electronic equipment for implementing steps based on a differential performance boundary test method;

[0133] The processor runs a program, and when the program runs, it executes the steps of the differential performance boundary test method based on the data output from the electronic device.

[0134] Storage medium for storing programs that, when run, execute steps of a differential performance boundary test method based on data output from electronic devices.

[0135] The communication bus mentioned in the above electronic devices can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. This communication bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, only one thick line is used to represent it in the diagram, but this does not mean that there is only one bus or one type of bus.

[0136] The electronic device comprises a hardware layer, an operating system layer running on top of the hardware layer, and an application layer running on the operating system. The hardware layer includes hardware such as a central processing unit (CPU), a memory management unit (MMU), and memory. The operating system can be any one or more computer operating systems that control the electronic device through processes, such as Linux, Unix, Android, iOS, or Windows. Furthermore, in this embodiment of the invention, the electronic device can be a smartphone, tablet computer, or other handheld device, or a desktop computer, portable computer, or other electronic device; there is no particular limitation in this embodiment.

[0137] In this embodiment of the invention, the executing entity for electronic device control can be an electronic device itself, or a functional module within an electronic device capable of calling and executing a program. The electronic device can obtain the firmware corresponding to the storage medium. This firmware is provided by the supplier, and different storage media may have the same or different firmware; no limitation is made here. After obtaining the firmware corresponding to the storage medium, the electronic device can write this firmware into the storage medium; specifically, it burns the firmware corresponding to the storage medium into the storage medium. The process of burning the firmware into the storage medium can be implemented using existing technology, and will not be elaborated upon in this embodiment of the invention.

[0138] Electronic devices can also obtain reset commands corresponding to the storage media. The reset commands corresponding to the storage media are provided by the supplier. The reset commands corresponding to different storage media can be the same or different, and no restrictions are imposed here.

[0139] At this time, the storage medium of the electronic device is a storage medium on which the corresponding firmware has been written. The electronic device can respond to the reset command corresponding to the storage medium on which the corresponding firmware has been written, thereby resetting the storage medium on which the corresponding firmware has been written according to the reset command. The process of resetting the storage medium according to the reset command can be implemented by existing technology and will not be described in detail in this embodiment of the invention.

[0140] For ease of description, the above devices are described separately by function as various units and modules. Of course, in implementing this application, the functions of each unit and module can be implemented in one or more software and / or hardware.

[0141] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the meaning consistent with their meaning in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined.

[0142] For the sake of simplicity, the method embodiments are described as a series of actions. However, those skilled in the art should understand that the embodiments of the present invention are not limited to the described order of actions, because according to the embodiments of the present invention, some steps can be performed in other orders or simultaneously. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to the embodiments of the present invention.

[0143] As can be seen from the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware platforms. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments of this application.

[0144] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method of testing differential performance boundaries, the method comprising: The differential performance boundary test method includes: Extract failure precursor factors from multimodal sensing data during real-time differential testing; In response to the failure precursor factor exceeding a preset failure threshold, the failure evolution path of the differential is determined based on the multimodal sensing data; Determine the active intervention strategy corresponding to the failure evolution path; The active intervention strategy is executed to determine the results of the differential performance boundary test.

2. The differential performance boundary test method according to claim 1, characterized in that, The extraction of failure precursor factors from the real-time multimodal sensing data during differential testing includes: Data from multiple types of sensors deployed during the differential bench test are collected in real time to obtain multimodal sensing data of the differential bench test; The encoder and decoder structures of the pre-trained autoencoder network are used to extract multimodal features from the multimodal sensing data. Based on the multimodal features, the error of the differential is reconstructed to generate a normalized failure precursor factor.

3. The differential performance boundary test method according to claim 2, characterized in that, Before determining the failure evolution path of the differential based on the multimodal sensing data, the method further includes: In response to the failure precursor factor exceeding a preset abnormal threshold but not exceeding the failure critical threshold, the frequency of collecting the multimodal sensing data is increased.

4. The differential performance boundary test method according to claim 1, characterized in that, Determining the failure evolution path of the differential based on the multimodal sensing data includes: The differential equations describing physical laws in the differential performance boundary test process are incorporated into the physical information neural network to determine the physical information neural network failure evolution inference model. Based on the physical information neural network failure evolution inference model, the multimodal sensing data is inferred to determine at least one failure evolution path; Each failure evolution path corresponds to a failure probability time series.

5. The differential performance boundary test method according to claim 4, characterized in that, The determination of the active intervention strategy corresponding to the failure evolution path includes: For each failure evolution path, the failure evolution path and its corresponding failure probability time series are determined as a set of failure evolution path data, resulting in multiple sets of failure evolution path data. Multiple sets of failure evolution path data are input into a deep reinforcement learning decision model to generate the optimal active intervention strategy with the goal of minimizing failure risk and maximizing the value of experimental data.

6. The differential performance boundary test method according to claim 5, characterized in that, The optimal proactive intervention strategy, aimed at minimizing failure risk and maximizing the value of experimental data, includes: With the goal of minimizing failure risk and maximizing the value of experimental data, the expected return of each set of failure evolution path data is evaluated on a millisecond timescale. Determine the optimal expected return, and in at least one of the failure evolution paths, determine the first failure evolution path corresponding to the optimal expected return; Determine the active intervention strategy corresponding to the first failure evolution path.

7. The differential performance boundary test method according to claim 1, characterized in that, The execution of the active intervention strategy to determine the differential performance boundary test results includes: The test bench controls the differential to execute the corresponding actions of the active intervention strategy to intervene in the failure of the differential until the failure precursor factor is less than or equal to the failure critical threshold. The differential is then tested using the test parameters following the implementation of the active intervention strategy to determine the performance boundary test results of the differential.

8. A differential performance boundary test device, characterized in that, The differential performance boundary test device includes: The failure factor extraction module is used to extract failure precursor factors in real time from the multimodal sensing data during the differential test process; The failure evolution path determination module is used to determine the failure evolution path of the differential based on the multimodal sensing data in response to the failure precursor factor exceeding a preset failure critical threshold. An intervention strategy determination module is used to determine the active intervention strategy corresponding to the failure evolution path; The test result determination module is used to execute the active intervention strategy and determine the performance boundary test results of the differential.

9. A computer-readable storage medium, characterized in that, The device stores a computer program executable by an electronic device, which, when run on the electronic device, causes the electronic device to perform the steps of the differential performance boundary test method as described in any one of claims 1 to 7.

10. A vehicle, characterized in that, include: An electronic device for implementing the steps of the differential performance boundary test method as described in any one of claims 1 to 7; A processor that runs a program that, when the program is running, performs the steps of the differential performance boundary test method as described in any one of claims 1 to 7 from data output by the electronic device. A storage medium for storing a program that, when run, performs the steps of the differential performance boundary test method as described in any one of claims 1 to 7 on data output from an electronic device.

Images