Automobile brake pad wear resistance detection method and system based on knowledge graph

Abstract

The application discloses a knowledge graph-based automobile brake pad wear resistance detection method and system, and belongs to the technical field of automobile part detection. The formula batch, pressing temperature curve, ablation layer thickness, porosity, wheel position of a to-be-detected brake pad, brake pedal opening, disc temperature and environmental humidity in a continuous downhill process in a mountain area after the brake pad is taken out of a refrigerator are obtained, and a knowledge graph with material parameters, process parameters, test working conditions, wear results and failure types as entity nodes is constructed. Similar historical wear samples are searched according to the knowledge graph, and a cross-wheel-position implicit wear propagation path is generated. The condensate retention coefficient and the resin layer peeling probability are calculated based on the path, and an unconventional wear resistance detection working condition is generated. An intermittent braking wear test is carried out in the working condition, the edge weight is corrected, and a scene wear resistance judgment value is obtained. The application improves the detection accuracy of the abnormal wear risk of the brake pad in a low-temperature and high-humidity continuous downhill scene.

Description

Technical Field

[0001] This invention relates to the field of automotive parts testing technology, specifically to a knowledge graph-based method and system for testing the wear resistance of automotive brake pads. Background Technology

[0002] Current methods for testing the wear resistance of automotive brake pads typically employ bench wear tests under fixed speed, fixed load, or preset temperature conditions. Brake pad quality is determined by the amount of wear, coefficient of friction, and post-test surface condition. While this method can evaluate the basic wear resistance of brake pads under standard operating conditions, the testing conditions are often based on manual experience or general standards, failing to fully consider the correlation between brake pad formulation batches, pressing temperature profiles, ablation layer thickness, porosity, wheel mounting location, and the actual operating environment of the vehicle.

[0003] In refrigerated transport vehicles, condensation easily forms on the surfaces of brake pads and discs after the vehicle leaves the low-temperature environment of a cold storage facility. When the vehicle then enters a continuous downhill section in a mountainous area, the driver frequently brakes intermittently, causing the brake disc temperature to rise rapidly while the ambient humidity remains high. This results in alternating evaporation and retention of moisture in the ablation layer, friction film, and internal pores of the brake pads. Especially for brake pads from the same batch but installed on different wheel positions, the axle load distribution, disc temperature differences, and wheel end heat dissipation conditions can lead to a rapid expansion of the friction film damage area on one side of the wheel position and an abnormally high wear rate per unit braking energy. This problem has a low probability of occurrence and is difficult to cover with routine spot checks, but once it occurs, it can cause rapid localized thinning of the brake pads, accelerated uneven wear, and even affect braking stability on long downhill sections.

[0004] Existing testing methods typically record material parameters, process parameters, test conditions, wear results, and failure types separately, lacking a mechanism for unified correlation and analysis of these data. This makes it difficult to identify cross-wheel wear patterns related to the brake pads under test from historical wear samples, and also fails to automatically generate wear resistance test conditions that more closely resemble the target scenario based on factors such as condensation retention and resin layer peeling. Therefore, existing technologies suffer from insufficient identification of abnormal brake pad wear under intermittent braking scenarios during continuous downhill driving in low-temperature, high-humidity conditions, and low matching degree between test results and actual vehicle wear resistance risks. Summary of the Invention

[0005] The purpose of this invention is to provide a knowledge graph-based method and system for testing the wear resistance of automotive brake pads, in order to address the shortcomings in the prior art.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a knowledge graph-based method for testing the wear resistance of automotive brake pads, comprising: The formula batch, pressing temperature curve, ablation layer thickness, porosity, wheel position of the brake pad to be tested, and brake pedal opening, disc temperature and ambient humidity of the vehicle after leaving the cold storage and going downhill in the mountainous area are obtained; the obtained data are mapped into a knowledge graph with material parameters, process parameters, test conditions, wear results and failure types as entity nodes. Based on the knowledge graph, historical wear samples with relevant conditions to the brake pad under test are retrieved, and a cross-wheel position hidden wear propagation path is generated; based on the propagation path, the condensate retention coefficient and resin layer peeling probability are calculated, and unconventional wear resistance test conditions are generated accordingly. Intermittent braking wear test is conducted under the unconventional wear test conditions to collect the wear amount per unit braking energy and the damaged area of ​​the friction film; the collected results are backfilled into the propagation path and the edge weights are corrected to obtain the scene wear resistance judgment value; when the scene wear resistance judgment value exceeds the preset threshold, it is determined that the brake pad under test does not meet the target scene wear resistance requirements.

[0007] Preferably, determining the continuous braking event segment includes: Align the brake pedal opening, disc temperature, and ambient humidity along the same time axis, and calculate the slope of disc temperature change and the slope of ambient humidity change respectively. Based on the non-braking reference segment, the effective threshold of the pedal and the disc temperature rise threshold are determined, and continuous data that simultaneously meets the conditions of effective pedal and disc temperature rise are determined as continuous braking event segments.

[0008] Preferably, the relationship edges of the knowledge graph are established through the following steps: In a continuous braking event segment, identify the disc temperature rise segment and the humidity fall segment, and calculate the overlap interval between the two; Using the start time of the overlapping interval as the trigger time of the relation edge, the material parameter node, process parameter node and test condition node are connected sequentially; The wear amount per unit braking energy, the damaged area of ​​the friction film, and the failure type are linked to the endpoints of the relation edges to form a knowledge graph.

[0009] Preferably, historical wear samples that exhibit conditions related to the brake pad under test are retrieved, including: Using the formula batch, ablation layer thickness, porosity and pressing temperature curve characteristics of the brake pads to be tested as entry nodes, historical wear samples with the same formula batch are first screened. The material process similarity distance is calculated for the screened historical wear samples. The material process similarity distance is obtained by normalizing the difference in ablation layer thickness, porosity, maximum slope difference in the heating stage, average temperature difference in the heat preservation stage, and heat preservation time difference. Samples with a material and process similarity distance no greater than the median material and process similarity distance of historical samples in the same batch were selected as historical wear samples related to the brake pads under test.

[0010] Preferably, generating a cross-wheel position latent wear propagation path includes: The relevant historical wear samples were grouped according to the wheel position of the vehicle, and the average wear per unit braking energy and the average friction film damage area were calculated for each wheel position. Calculate the wear difference between any two wheel positions. When the wear difference is greater than the average difference of the same batch of samples, mark the corresponding wheel position relationship as an implicit wheel position association edge. The direction of the hidden wheel position associated edge is determined based on the direction of the increased wear result, and it is connected to the historical failure type node to obtain the hidden wear propagation path across wheel positions.

[0011] Preferably, the calculation of the condensate retention coefficient includes: Read the overlapping section of the disk temperature rise segment and the humidity fall segment from the propagation path of hidden wear across the wheel position, and calculate the duration of the overlapping section and the amount of environmental humidity decrease within the overlapping section; The duration of the overlapping interval, the decrease in ambient humidity, and the porosity were normalized based on the 95th percentile of historical wear samples from the same batch of formulations. The condensate retention coefficient was obtained by weighting the duration, the decrease in humidity, and the porosity.

[0012] Preferably, calculating the resin layer peeling probability includes: Convert the wear result difference value of the implicit wheel position associated edge into the implicit wheel position associated edge weight; The average friction film damage area is read from the end wheel position of the wear increase direction of the hidden wear propagation path across the wheel position, and combined with the ablation layer thickness of the brake pad under test, the normalized friction film damage area and the normalized ablation layer thinness are obtained respectively. By inputting the implicit wheel position associated edge weights, the normalized friction film damage area, and the normalized ablation layer weakness into the logic function, the resin layer peeling probability is obtained.

[0013] Preferably, generating unconventional wear resistance testing conditions includes: The humidity retention time is obtained by extending the median time from when the vehicle leaves the cold storage facility to the start of the first continuous braking event based on the condensate retention coefficient. The braking interval is obtained by shortening the median of the interval between adjacent braking start points within a historical continuous braking event segment based on the resin layer peeling probability. Based on the condensate retention coefficient and resin layer peeling probability, the median of the highest historical plate temperature is mapped to the 95th percentile of the highest historical plate temperature to obtain the upper limit of the plate temperature.

[0014] Preferably, the scene wear resistance determination value is obtained, including: According to the humidity holding time, braking interval and disc temperature limit, the bench test is divided into multiple equal braking energy intervals, and the wear amount per unit braking energy and the increase in friction film damage area are calculated after each equal braking energy interval. The intermittent segment in which the wear of unit braking energy increases beyond the weighing error and the increase in the damaged area of ​​the friction film increases beyond the image recognition error is defined as the propagation path backfill segment. The implicit wheel position associated edge weights are corrected based on the degree of deviation between the unit braking energy wear amount and the increase in friction film damage area of ​​the backfill section of the propagation path and the historical wear results, and the scenario wear resistance judgment value is calculated from the corrected implicit wheel position associated edge weights.

[0015] This invention also provides a knowledge graph-based automotive brake pad wear resistance testing system, comprising: a data acquisition module, a testing condition generation module, and a brake pad testing output module.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention differs from existing technologies that rely solely on fixed loads, rotational speeds, or temperatures for general wear tests. Instead, it incorporates the brake pad's batch composition, pressing temperature curve, ablation layer thickness, porosity, wheel position, and brake pedal opening, disc temperature, and ambient humidity during a continuous downhill descent in mountainous terrain after the vehicle leaves cold storage as input for testing. The overlapping area between the disc temperature rise and humidity decrease segments serves as the triggering basis for knowledge graph relationship edges. This core technology doesn't simply record vehicle operating conditions; instead, it transforms the actual wear-inducing factor—brake temperature rise after low temperature and high humidity leading to moisture evaporation, friction film disturbance, and material pore moisture absorption—into a condition triggering path that can be linked to material parameters, process parameters, and wear results. This solves the problem that existing testing methods cannot cover the risk of abnormal uneven brake pad wear, rapid localized thinning, and friction film peeling when cold chain vehicles immediately enter long downhill mountainous areas after leaving storage. It establishes a direct correspondence between testing conditions and actual failure scenarios, improving the scenario-specificity of wear resistance testing.

[0017] Furthermore, this invention screens historical wear samples with similar formulations and processes by using material and process similarity distances. It then calculates the wear amount per unit braking energy and the difference in friction film damage area based on the wheel position, forming a hidden wear propagation path across wheel positions. Based on this path, it calculates the condensate retention coefficient and resin layer peeling probability, thereby generating humidity retention time, braking interval, and upper limit of disc temperature. This core technology reverses the historical failure path into bench test boundaries, ensuring that unconventional wear resistance testing conditions are not artificially set but derived from the same source materials, actual wheel position differences, and wet-heat coupled wear characteristics. By collecting wear amount and friction film damage increments during equal braking energy intervals and backfilling and correcting the hidden wheel position related edge weights, this invention can achieve closed-loop correction between current bench test results and historical abnormal wear paths, thereby improving the accuracy and repeatability of wear risk assessment for low-frequency but high-risk scenarios. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the method of the present invention.

[0020] Figure 2 This is a flowchart of the system modules of the present invention.

[0021] Figure 3 Determine the flowchart for the continuous braking event segment.

[0022] Figure 4 Build a flowchart for the relationship edges of the knowledge graph.

[0023] Figure 5 A flowchart is generated to show the propagation path of hidden wear across wheel positions.

[0024] Figure 6 Flowchart for calculating the wear resistance value for a given scenario. Detailed Implementation

[0025] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. 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.

[0026] Example 1, please refer to Figure 1 and Figure 3 and Figure 4 and Figure 5 and Figure 6 As shown in this embodiment, the knowledge graph-based method for testing the wear resistance of automotive brake pads includes: The formula batch, pressing temperature curve, ablation layer thickness, porosity, wheel position of the brake pad to be tested, and brake pedal opening, disc temperature and ambient humidity of the vehicle after leaving the cold storage and going downhill in the mountainous area are obtained; the obtained data are mapped into a knowledge graph with material parameters, process parameters, test conditions, wear results and failure types as entity nodes. In this embodiment, after the brake pad under test is manufactured, a unique identification code is set on its back plate. This unique identification code corresponds one-to-one with the formula batch and is stored in the production data table, which records the friction material formula number, resin material batch number, reinforcing fiber batch number, filler batch number, and mixing batch number. After reading the unique identification code, the testing and computing equipment retrieves the formula batch corresponding to the brake pad under test according to the unique identification code.

[0027] After hot pressing is completed, the pressing temperature curve of the brake pad under test is read from the data log file of the hot pressing equipment. The pressing temperature curve includes the upper mold temperature, lower mold temperature, heating time, holding time, and cooling time, with a sampling interval of 1 second to 10 seconds. At the start of each pressing, the mold cavity number and the unique identification code of the brake pad are written into the production data table to ensure that the pressing temperature curve corresponds to the brake pad under test.

[0028] The ablation layer thickness and porosity of samples from the same formulation and pressing batch as the brake pads under test were measured. The ablation layer thickness was obtained through cross-sectional microscopic images along the thickness direction of the samples. Three measurement points were selected along the length of the friction material surface to identify the grayscale boundary between the surface ablation layer and the substrate material, and the thickness values ​​were obtained at each point. , , ablation layer thickness Calculate using the following formula: ;in, The ablation layer thickness is expressed in mm.

[0029] Porosity was obtained using Archimedes' drainage method. The dried mass of the samples was measured. , saturated mass of immersion solution and suspended mass Porosity Calculate using the following formula: ;in, Porosity For drying quality, For the saturated mass of the immersion solution, This refers to the suspended mass.

[0030] When brake pads are installed on a vehicle, their unique identification code is read, and the vehicle identification code and the installation wheel position are recorded. The installation wheel position is represented by fixed fields: left front wheel, right front wheel, left rear wheel, and right rear wheel. This establishes a one-to-one correspondence between the unique identification code, the vehicle identification code, and the installation wheel position.

[0031] When vehicles are performing cold chain transportation tasks, brake pedal opening, disc temperature, and ambient humidity are collected during the continuous downhill descent in mountainous areas after leaving the cold storage. Brake pedal opening is read from vehicle bus data, disc temperature is obtained from infrared thermometers or thermocouples located near the brake discs of the corresponding wheels, and ambient humidity is obtained from temperature and humidity sensors located inside the wheel wells. The sampling interval is 100 ms to 1 s, a range used to cover the rapid changes in pedal opening and disc temperature during a single braking maneuver.

[0032] The conditions for continuous downhill driving in mountainous areas after a vehicle leaves the cold storage are determined as follows: the vehicle must remain parked in the cold storage area for at least 30 minutes, and within 20 minutes of leaving the cold storage, it must enter a section of road with an average gradient of at least 4% and a continuous downhill distance of at least 2 km. The 30-minute parking time is determined by testing the brake pad backplate temperature to be close to the ambient temperature of the cold storage, specifically the time required for the backplate temperature to be less than or equal to 2°C, rounded up. The 20-minute parking time is determined by the duration during which condensation may still form on the wheel ends after the vehicle leaves the cold storage. The average gradient of 4% and the continuous distance of 2 km are determined by statistical values ​​of the gradient and distance of continuous braking sections in the target mountainous transportation route, respectively, and the 75th percentile is used to cover downhill scenarios with higher braking intensity than in conventional urban areas.

[0033] Subsequently, brake pedal opening, disc temperature, and ambient humidity were arranged along the same time axis. If the sampling times of the three types of data were inconsistent, the maximum sampling interval was used as the reference time interval. Then, perform linear interpolation on the remaining data. For any variable... At any moment The interpolation is calculated using the following formula: ;in, , and These are the data values ​​from two adjacent sampled times.

[0034] Plate temperature change slope and the slope of changes in ambient humidity Calculate according to the following formulas: , ;in, For a moment The plate temperature, For a moment The ambient humidity The unit is ℃ / s. The unit is %RH / s.

[0035] Calculate the average brake pedal opening during the non-braking reference section before the downhill slope. and standard deviation effective threshold of pedal Determine using the following formula: Within the same non-braking reference segment, calculate the average slope of the disc temperature change. and standard deviation plate temperature rise threshold Determine using the following formula: The reason for using three times the standard deviation is to eliminate fluctuations caused by sensor noise, natural heat dissipation, and vehicle airflow, so that the identified disc temperature rise corresponds to the braking energy input.

[0036] when and When two sampling points are satisfied consecutively, the moment when the first condition is met is determined as the starting point of the continuous braking event segment; when and When two sampling points are satisfied consecutively, the moment when the condition is met is determined as the end point of the continuous braking event segment. The data between the start point and the end point constitute the continuous braking event segment. .

[0037] For each continuous braking event segment Calculate the average pedal opening. , maximum plate temperature Cumulative increase in disk temperature and cumulative decrease in ambient humidity : , , , ;in, This represents the number of sampling points within a continuous braking event segment.

[0038] Then, the continuous braking event segments Compared with the formula batch Characteristics of pressing temperature curve ablation layer thickness Porosity and wheel position Binding, generating batch wheel position working condition composite entities : Among them, the characteristics of the pressing temperature curve The parameters include the maximum slope during the heating phase, the average temperature during the holding phase, and the holding time, which are used to characterize the impact of the hot pressing process on the brake pad material structure.

[0039] When establishing the relation edges of the knowledge graph, start from the continuous braking event fragments. The process involves defining the temperature rise phase and the humidity decrease phase. The temperature rise phase is designed to meet the following requirements: The continuous time interval; the humidity decrease period is to satisfy The continuous time interval. Humidity decrease threshold. Determine using the following formula: ;in, Within the non-braking reference section The average value, Within the non-braking reference section The standard deviation of the value. This threshold is used to exclude natural fluctuations in ambient humidity, ensuring that the decrease in humidity corresponds to the evaporation process caused by the increase in dish temperature.

[0040] Let the temperature rise segment be The humidity decreases during the period. The overlapping area of ​​the two Determine using the following formula: The overlapping interval is determined when the following formula is satisfied: ; Overlapping intervals The starting time is used as the trigger time for the relation edge. The material parameter node, process parameter node and test condition node are connected in sequence to form a relation edge with the direction of "material parameter node → process parameter node → test condition node".

[0041] Finally, it will be combined with this composite entity The corresponding wear result node and failure type node are attached to the endpoint of the relationship edge. Wear results include wear per unit braking energy. and the area of ​​friction film damage In the bench reproduction test, the braking energy Calculate using the following formula: ;in, for The braking torque applied to the wheel brake disc at a given moment, not the total braking torque of the entire vehicle. for Brake disc angular velocity at any given moment. Wear per unit braking energy. Calculate using the following formula: ;in, The difference in brake pad quality before and after the test resulted in an area of ​​friction film damage. Calculations were made based on images of the friction surface after the experiment. Before shooting, images with a side length of [missing information] were used. The calibration block determines the area of ​​a single pixel. : ;in, This represents the number of pixels corresponding to the calibration block in the image. The number of pixels in the damaged area is denoted as... Area of ​​friction film damage Calculate using the following formula: The failure type nodes are represented by fixed codes: F1 is uneven wear, F2 is groove wear, F3 is friction film peeling, and F4 is thermal cracking.

[0042] Thus, each relation edge is triggered by the overlapping interval of the plate temperature rise segment and the humidity fall segment, and ultimately points to the determined wear result and failure type, forming a brake pad wear resistance knowledge graph with working condition triggering direction.

[0043] This invention quantifies the actual energy input to the brake pair at each sampling moment during intermittent braking into braking energy by integrating the braking torque and brake disc angular velocity over time. This energy is then used to calculate the wear per unit of braking energy, thus avoiding errors caused by evaluating wear resistance solely based on the number of braking cycles, braking time, or total wear mass. Furthermore, by clearly defining the angular velocity unit conversion, the corresponding wheel position braking torque, and the positive values ​​for braking energy, the invention ensures the comparability of wear data under different vehicle, wheel, and bench conditions. This allows the wear result nodes in the subsequent knowledge graph to accurately reflect the energy-related wear characteristics of the brake pad under test in the target scenario, improving the match between wear resistance test results and actual vehicle wear risks.

[0044] Based on the knowledge graph, historical wear samples with relevant conditions to the brake pad under test are retrieved to generate a hidden wear propagation path across wheel positions; the condensate retention coefficient and resin layer peeling probability are calculated based on the propagation path, and unconventional wear resistance test conditions are generated accordingly.

[0045] In this embodiment, the formula batch corresponding to the brake pad to be tested is used. ablation layer thickness Porosity and pressing temperature curve characteristics As an entry node. Characteristics of the pressing temperature curve. Including the maximum slope during the heating phase Average temperature during the heat preservation stage and heat preservation time In the knowledge graph, for each historical wear sample... Read its formula batch ablation layer thickness Porosity Maximum slope during the heating phase Average temperature during the heat preservation stage and heat preservation time .when When, retain the historical wear sample; when In this case, the historical wear sample will not be included in the current cross-wheel position analysis.

[0046] For historical wear samples with the same formulation batch, calculate the material and process similarity distance. : ;in, , , , and These are the differences between the maximum and minimum values ​​of the corresponding parameters in historical wear samples of the same formula batch. When any difference is 0, it is replaced with the minimum resolution of the corresponding testing instrument. The above normalization calculation is used to ensure that the ablation layer thickness, porosity, maximum slope during the heating stage, average temperature during the holding stage, and holding time can all participate in similarity judgment under different dimensions.

[0047] All Arrange in ascending order, and take... No more than all of the same formula batch The median historical wear sample serves as a sample of historical wear data with similar material and process parameters. The median is calculated from historical data of the same batch obtained in this search, without using a fixed empirical threshold, and is used to remove samples that deviate from the normal manufacturing conditions of the same batch.

[0048] Then, the retrieved historical wear samples are sorted according to the wheel position. The wheels are grouped into four positions: left front wheel, right front wheel, left rear wheel, and right rear wheel. For each wheel position... Extract the unit braking energy wear amount of each historical wear sample at this wheel position. and the area of ​​friction film damage Calculate the average wear per unit braking energy at that wheel position. and average friction film damage area : , ;in, For loading wheel position The number of historical wear samples below For the first position of this round The wear rate per unit braking energy of a sample. For the first position of this round The area of ​​friction film damage in each sample. When a certain wheel position... In this case, the wheel position is not included in the calculation of subsequent wheel position combinations.

[0049] Furthermore, calculate any two loading wheel positions. and Difference in wear results between : ;in, This is the difference between the maximum and minimum wear values ​​per unit braking energy in the retrieved historical wear samples. This is the difference between the maximum and minimum areas of damage to the friction film. When... At that time, the mass measurement accuracy is divided by the corresponding median braking energy as the value. ;when At that time, the area of ​​a single pixel obtained from image calibration is taken as... This difference value includes both mass wear difference and surface damage difference, and is used to avoid judging wheel position abnormalities solely based on wear quality.

[0050] Average difference of samples in the same batch Calculate using the following formula: ;in, This refers to the number of wheel position combinations involved in the calculation; 4 loading wheel positions can form a maximum of 6 wheel position combinations. When a certain wheel position relationship satisfies... When this occurs, the wheel position relationship is marked as an implicit wheel position associated edge. If This indicates that the wheel position combination involved in the calculation does not show wear differences, and in this case, no implicit wheel position associated edges are generated (i.e., the edge weight is considered to be 0 or no connection is established).

[0051] Implicit wheel position associated edge weight Calculate using the following formula: ;in, The value range is 0 to 1. This formula converts wheel wear differences into dimensionless edge weights, enabling comparison of wheel wear correlation strengths across different batches or wear levels. (Note: If...) ,but It is also 0, at which point it is defined. ).

[0052] The path direction is determined as follows. First, calculate the wheel position. Relative to wheel position normalized contribution of wear Normalized contribution of damaged area : ;when and At that time, the path direction is ;when and At that time, the path direction is When the signs of the two are opposite, compare and The direction of increased wheel wear corresponding to the item with the larger absolute value is taken as the path direction. Thus, starting from the entry node of the brake pad under test, the marked implicit wheel position associated edges are connected, and these implicit wheel position associated edges are pointed to the failure type nodes recorded in the corresponding historical wear samples, forming a cross-wheel position implicit wear propagation path.

[0053] Subsequently, the implicit wheel position associated edge weights were read from the implicit wear propagation path across wheel positions. The overlapping area between the temperature rise and humidity fall of the disk. Porosity and ablation layer thickness Overlapping intervals are denoted as: ;in, The starting time of the overlapping interval. This is the termination time of the overlapping interval. The duration of the overlapping interval. Calculate using the following formula: ; Decrease in ambient humidity within overlapping regions Calculate using the following formula: ;when At that time, take .

[0054] The duration of the overlapping interval, the decrease in ambient humidity, and the porosity were normalized. , , ;in, , and These are the 95th percentiles for the duration of overlapping intervals, the decrease in ambient humidity, and porosity among historical wear samples from the same formulation batch. When there are fewer than 20 historical samples from the same formulation batch, all historical samples under the same friction material formulation number are used for calculation; when the obtained percentile is 0, the minimum non-zero measurement resolution of the corresponding data is used instead. The 95th percentile is used to represent the level of high-intensity damp heat exposure that has occurred and to reduce the impact of a single anomaly maximum value on the testing conditions.

[0055] Condensate retention coefficient Calculate using the following formula: ;in, The value ranges from 0 to 1. The duration has a weight of 0.4, which is used to reflect the effect of condensate residence time on pore adsorption and subsequent vaporization impact; the humidity decrease and porosity both have weights of 0.3, which are used to reflect the evaporation intensity and the material's water holding capacity, respectively.

[0056] Further calculations were made to determine the probability of resin layer peeling. The area of ​​friction film damage was also calculated. The average friction film damage area is taken from the wheel position at the endpoint of the wear increase direction in the hidden wear propagation path across wheel positions, i.e., if the path direction is... ,but If the path direction is ,but Normalized friction film damage area and normalized ablation layer thinness Calculate according to the following formulas: , ;in, This represents the 95th percentile of the area of ​​friction film damage in historical wear samples from the same formulation batch. This represents the 95th percentile of the ablation layer thickness in historical wear samples from the same formulation batch. At that time, the area of ​​a single pixel obtained from image calibration is taken as... ;when At that time, the thickness resolution of the microscopic measuring equipment was taken as... The thinner the ablation layer, the weaker the surface carbonization protection; therefore, the following methods are used. Characterizes the thinness of the ablation layer.

[0057] In this embodiment, the resin layer peeling probability Calculate using the following formula: ;in, This represents the natural exponential function. The formula maps the wear difference across wheel positions, the area of ​​friction film damage, and the thinning of the ablation layer to probability values ​​between 0 and 1. Coefficients 1.2, 1.0, and 0.8 correspond to the contributions of wear difference across wheel positions, friction film damage, and ablation layer thinning to the peeling risk, respectively. A constant of 1.5 is used to ensure that the peeling probability is close to 0.5 when the three inputs are at a moderate level. When there are at least 50 historical samples labeled with friction film peeling failure types, the above coefficients are obtained through logistic regression fitting.

[0058] Finally, the condensate retention coefficient and resin layer peeling probability Mapped to humidity holding time in bench tests Braking interval Upper limit of plate temperature : ;in, The median time from when a vehicle leaves the cold storage to the start of the first continuous braking event in the historical sample: ;in, This is the median of the interval between two consecutive braking start points within a historical continuous braking event segment. A coefficient of 0.5 allows the braking interval to be shortened to a maximum of 50% of the historical median, used to simulate frequent intermittent braking while avoiding excessively short intervals that would cause the test bench conditions to deviate from the actual braking rhythm of the vehicle.

[0059] Upper plate temperature Calculate using the following formula: ;in, This represents the median of the highest disc temperatures across historical consecutive braking event segments from the same batch of formulations. This represents the 95th percentile of the highest disc temperature in a historical sequence of consecutive braking events from the same formulation batch. Because... and The maximum value of all of them is 1, therefore The maximum value is 1, thus ensuring Limited to and Between these points, avoid exceeding the upper limit of the disk temperature under historical high-intensity operating conditions.

[0060] Therefore, the humidity retention time is obtained. Braking interval Upper limit of plate temperature The unconventional wear resistance testing conditions are jointly defined and correspond to the wet heat retention characteristics and resin layer peeling risk in the cross-wheel position latent wear propagation path.

[0061] This invention first screens similar historical wear samples based on formulation batches, ablation layer thickness, porosity, and pressing temperature curve characteristics. Then, it calculates the differences in wear per unit braking energy and friction film damage area between different wheel positions. This allows for the identification of cross-wheel position abnormal wear correlations that are difficult to detect under conventional testing conditions with the same material and similar process. Furthermore, this correlation is converted into implicit wheel position correlation edge weights, and combined with the overlapping interval of disc temperature rise and humidity decrease, porosity, and ablation layer thickness, the condensate retention coefficient and resin layer peeling probability are calculated. This makes the unconventional wear resistance testing conditions no longer dependent on manual experience settings, but rather derived from actual historical failure paths and target humid and hot conditions. The humidity holding time, braking interval, and disc temperature upper limit generated in this way can more closely resemble the low-frequency, high-risk scenario of continuous downhill driving in mountainous areas after leaving cold storage, improving the accuracy of brake pad wear resistance testing in identifying risks of uneven wear, friction film peeling, and rapid wear.

[0062] Intermittent braking wear test is conducted under the unconventional wear test conditions to collect the wear amount per unit braking energy and the damaged area of ​​the friction film; the collected results are backfilled into the propagation path and the edge weights are corrected to obtain the scene wear resistance judgment value; when the scene wear resistance judgment value exceeds the preset threshold, it is determined that the brake pad under test does not meet the target scene wear resistance requirements.

[0063] In this embodiment, the humidity holding time is based on the unconventional abrasion resistance testing conditions. Braking interval Upper limit of plate temperature Conduct intermittent brake wear test. Before the test, place the brake pads and matching brake discs under test in an environment with a relative humidity of [missing information]. Maintain in the environment ,in Take the median of the initial humidity in the overlapping section of the disk temperature rise segment and humidity fall segment in the propagation path of hidden wear across the wheel position; when there is only one propagation path, take the median of the humidity in that path. As This ensures that the initial humidity level of the test corresponds to the humid and hot state before braking after the vehicle leaves the cold storage.

[0064] The bench test was divided into multiple equal braking energy intervals. The target braking energy for each interval was denoted as... , The median braking energy of a single braking event is taken from the historical continuous braking event segment corresponding to the propagation path of latent wear across wheel positions. Total experimental energy. Take the median of the cumulative braking energy corresponding to the historical continuous downhill conditions along this propagation path, and the number of intermittent segments. Determine using the following formula: ;in, This indicates rounding up. The median is used because it comes from historical data under the same or similar material and process conditions, which can reduce the impact of a single sudden stop or abnormal sampling on the test intensity.

[0065] No. Actual braking energy of each intermittent segment Calculate using the following formula: ;in, For the first The number of sampling points within each interval. If the sampled data is the brake disc speed... The unit is Then the angular velocity is converted using the following formula: ; when At that time, the first... There are several intervals. If the plate temperature reaches [a certain level] within that interval... and Then stop braking and allow the disc temperature to drop naturally. Then continue in that interval until... .in The temperature measurement error should be set to more than five times the upper limit to prevent the plate temperature from repeatedly exceeding the upper limit.

[0066] After each braking energy interval ends, wait for the brake pads to cool down. The following steps involve weighing and photographing the friction surface. (Select) The reason is that this temperature is lower than the safe contact temperature commonly used during manual weighing and image capture, and it can reduce the impact of thermal expansion and contraction on quality and image boundary recognition. Quality difference of each interval Calculate using the following formula: ;in, For the first Brake pad mass before the start of each interval segment For the first Brake pad quality after each interval. Wear per unit braking energy during each intermittent segment Calculate using the following formula: ; The damaged area of ​​the friction film was calculated using images of the friction surface. Before the experiment, images of the intact friction film area were taken, and the average grayscale value of the intact friction film was obtained. and grayscale standard deviation Damaged grayscale threshold Determine using the following formula: ;No. After a few intervals, the grayscale value will be less than [a certain value]. The pixels with the following characteristics are selected as candidate damaged pixels. To eliminate dust and imaging noise, the candidate damaged pixels need to form connected regions, and the number of pixels in the connected regions must be no less than [a certain value]. : ;in, This represents the actual area corresponding to a single pixel. The 95th percentile of the area of ​​the pseudo-damaged connected region in repeatedly captured images of the intact friction film under the same shooting conditions is taken. The area of ​​friction film damage after each interval Calculate using the following formula: ;in, The number of pixels in the damaged area to meet the above conditions. Increment of the damaged area of ​​the friction film. Calculate using the following formula: ; The propagation path backfilling section is determined starting from the second intermittent segment. The quality measurement resolution is denoted as... Increased wear per unit braking energy as a criterion Determine using the following formula: ; Image area recognition error is denoted as When the first occurrence of the first When all intervals simultaneously satisfy the following formula, the first interval will be... The intermittent segments were identified as backfill segments in the propagation path: , The above judgment indicates that the wear rate of this intermittent segment not only exceeds the weighing error, but also that the friction film damage and expansion exceed the image recognition error. If none of the intermittent segments meet the above conditions, then the [missing information] segment will be [missing information]. Each intermittent segment serves as a backfill segment for the propagation path.

[0067] Let the backfill segment of the propagation path be the first... Each interval. The average unit braking energy wear amount is read from the historical wear results corresponding to this propagation path. and the increase in average friction film damage area The historical wear results are taken from historical samples of the wheel position at the end of the wear increase direction in this propagation path. Degree of deviation. and Calculate using the following formula: , ; in, This represents the range of wear per unit braking energy in this historical sample. This represents the range of the increase in the damaged area of ​​the friction film; when the range is 0... Pick , Pick .

[0068] Corrected implicit wheel position associated edge weights Calculate using the following formula: ;in, The implicit wheel position associated edge weights before correction. Coefficients of 0.5 and 0.5 ensure that abnormal mass wear and abnormal friction film damage contribute equally to the edge weight correction, avoiding a single indicator dominating the judgment.

[0069] Scene wear resistance judgment value Calculate using the following formula: ;in, This represents the 95th percentile of wear per unit braking energy in historical wear samples from the same batch of formulations. The 95th percentile of the increase in friction film damage area among historical wear samples of the same formulation batch; when the percentile is 0, respectively using and Replacement.

[0070] Among them, the judgment threshold Take the 95th percentile of the wear resistance judgment value of historical qualified sample scenarios; when there are fewer than 20 historical qualified samples... Take the maximum wear resistance value from historical qualified sample scenarios; when there are no historical qualified samples... .because The value ranges from 0 to 1, and is composed of the modified edge weight, the wear amount per unit braking energy, and the increase in the area of ​​friction film damage. A value of 0.8 indicates that at least one major risk item is at a high level, which is used to form a conservative judgment boundary.

[0071] If the following formula is met, the brake pad under test is determined to not meet the wear resistance requirements of the target scenario: ; When the following formula is satisfied, the brake pad under test is determined to meet the wear resistance requirements of the target scenario: ; Therefore, the wear amount per unit braking energy and the increase in friction film damage area collected by the intermittent braking wear test can be backfilled into the hidden wear propagation path across the wheel position, and the scenario wear resistance judgment value can be obtained through edge weight correction, so that the final judgment result simultaneously reflects the correlation between actual energy wear, friction film damage expansion and historical abnormal wear across the wheel position.

[0072] This invention further divides unconventional wear resistance testing conditions into multiple equal braking energy intervals, enabling wear results under different braking frequencies and durations to be compared on the same energy scale, avoiding misjudgments caused by simply evaluating based on time or frequency. Simultaneously, after each interval, the wear amount per unit braking energy and the increase in friction film damage area are collected synchronously. This allows for the identification of abnormal nodes where the wear rate increases and friction film damage expands simultaneously. These nodes are then used as backfill segments in the propagation path to correct the implicit wheel position association edge weights, ensuring that historical cross-wheel position abnormal associations in the knowledge graph are corrected by the current bench test results. The final scenario wear resistance judgment value integrates historical failure paths, current energy wear intensity, and friction film damage expansion, thereby improving the accuracy and repeatability of judging abnormal wear risk of brake pads in low-frequency, high-risk scenarios such as continuous downhill driving in mountainous areas after leaving cold storage.

[0073] Example 2, please refer to Figure 2 As shown in this embodiment, the knowledge graph-based automotive brake pad wear resistance testing system includes: The data acquisition module acquires the formula batch, pressing temperature curve, ablation layer thickness, porosity, wheel position of the brake pads to be tested, and the brake pedal opening, disc temperature and ambient humidity of the vehicle after leaving the cold storage and going downhill in the mountainous area. The acquired data is mapped into a knowledge graph with material parameters, process parameters, test conditions, wear results and failure types as entity nodes. The test condition generation module retrieves historical wear samples that have relevant conditions to the brake pad under test based on the knowledge graph, generates a hidden wear propagation path across wheel positions, calculates the condensate retention coefficient and resin layer peeling probability based on the propagation path, and generates unconventional wear resistance test conditions accordingly. The brake pad detection output module performs intermittent brake wear tests under the unconventional wear resistance test conditions, collects the wear amount per unit braking energy and the damaged area of ​​the friction film; the collected results are backfilled into the propagation path and the edge weights are corrected to obtain the scene wear resistance judgment value, and when the scene wear resistance judgment value exceeds the preset threshold, it is determined that the brake pad under test does not meet the target scene wear resistance requirements.

[0074] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A knowledge graph-based method for testing the wear resistance of automotive brake pads, characterized in that, include: The formula batch, pressing temperature curve, ablation layer thickness, porosity, wheel position of the brake pad to be tested, and brake pedal opening, disc temperature and ambient humidity of the vehicle after leaving the cold storage and going downhill in the mountainous area are obtained; the obtained data are mapped into a knowledge graph with material parameters, process parameters, test conditions, wear results and failure types as entity nodes. Based on the knowledge graph, historical wear samples that are related to the brake pad under test are retrieved, and a cross-wheel position hidden wear propagation path is generated. The condensate retention coefficient and resin layer peeling probability are calculated based on the propagation path, and unconventional wear resistance test conditions are generated accordingly. Intermittent braking wear test was conducted under the unconventional wear resistance test conditions to collect the wear amount per unit braking energy and the damaged area of ​​the friction film. The collected results are backfilled into the propagation path and the edge weights are corrected to obtain the scene wear resistance judgment value. When the scene wear resistance judgment value exceeds the preset threshold, it is determined that the brake pad under test does not meet the target scene wear resistance requirements.

2. The method for testing the wear resistance of automotive brake pads based on knowledge graphs according to claim 1, characterized in that, Determine the segments of continuous braking events, including: Align the brake pedal opening, disc temperature, and ambient humidity along the same time axis, and calculate the slope of disc temperature change and the slope of ambient humidity change respectively. Based on the non-braking reference segment, the effective threshold of the pedal and the disc temperature rise threshold are determined, and continuous data that simultaneously meets the conditions of effective pedal and disc temperature rise are determined as continuous braking event segments.

3. The knowledge graph-based method for testing the wear resistance of automotive brake pads according to claim 2, characterized in that, The relationship edges of a knowledge graph are established through the following steps: In a continuous braking event segment, identify the disc temperature rise segment and the humidity fall segment, and calculate the overlap interval between the two; Using the start time of the overlapping interval as the trigger time of the relation edge, the material parameter node, process parameter node and test condition node are connected sequentially; The wear amount per unit braking energy, the damaged area of ​​the friction film, and the failure type are linked to the endpoints of the relation edges to form a knowledge graph.

4. The knowledge graph-based method for testing the wear resistance of automotive brake pads according to claim 1, characterized in that, Retrieve historical wear samples that are related to the conditions of the brake pad under test, including: Using the formula batch, ablation layer thickness, porosity and pressing temperature curve characteristics of the brake pads to be tested as entry nodes, historical wear samples with the same formula batch are first screened. The material process similarity distance is calculated for the screened historical wear samples. The material process similarity distance is obtained by normalizing the difference in ablation layer thickness, porosity, maximum slope difference in the heating stage, average temperature difference in the heat preservation stage, and heat preservation time difference. Samples with a material and process similarity distance no greater than the median material and process similarity distance of historical samples in the same batch were selected as historical wear samples related to the brake pads under test.

5. The knowledge graph-based method for testing the wear resistance of automotive brake pads according to claim 4, characterized in that, Generate a cross-wheel position latent wear propagation path, including: The relevant historical wear samples were grouped according to the wheel position of the vehicle, and the average wear per unit braking energy and the average friction film damage area were calculated for each wheel position. Calculate the wear result difference between any two wheel positions. When the wear result difference is greater than the average difference of the same batch of samples, mark the corresponding wheel position relationship as an implicit wheel position association edge. The direction of the hidden wheel position associated edge is determined based on the direction of the increased wear result, and it is connected to the historical failure type node to obtain the hidden wear propagation path across wheel positions.

6. The method for testing the wear resistance of automotive brake pads based on knowledge graphs according to claim 1, characterized in that, Calculate the condensate retention coefficient, including: Read the overlapping section of the disk temperature rise segment and the humidity fall segment from the propagation path of hidden wear across the wheel position, and calculate the duration of the overlapping section and the amount of environmental humidity decrease within the overlapping section; The duration of the overlapping interval, the decrease in ambient humidity, and the porosity were normalized based on the 95th percentile of historical wear samples from the same batch of formulations. The condensate retention coefficient was obtained by weighting the duration, the decrease in humidity, and the porosity.

7. The method for testing the wear resistance of automotive brake pads based on knowledge graphs according to claim 1, characterized in that, Calculating the resin layer peeling probability includes: Convert the wear result difference value of the implicit wheel position associated edge into the implicit wheel position associated edge weight; The average friction film damage area is read from the end wheel position of the wear increase direction of the hidden wear propagation path across the wheel position, and combined with the ablation layer thickness of the brake pad under test, the normalized friction film damage area and the normalized ablation layer thinness are obtained respectively. By inputting the implicit wheel position associated edge weight, the normalized friction film damage area, and the normalized ablation layer weakness into the logic function, the resin layer peeling probability is obtained.

8. The method for testing the wear resistance of automotive brake pads based on knowledge graphs according to claim 1, characterized in that, Generate unconventional wear resistance testing conditions, including: The humidity retention time is obtained by extending the median time from when the vehicle leaves the cold storage facility to the start of the first continuous braking event based on the condensate retention coefficient. The braking interval is obtained by shortening the median of the interval between adjacent braking start points within a historical continuous braking event segment based on the resin layer peeling probability. Based on the condensate retention coefficient and resin layer peeling probability, the median of the highest historical plate temperature is mapped to the 95th percentile of the highest historical plate temperature to obtain the upper limit of the plate temperature.

9. The method for testing the wear resistance of automotive brake pads based on knowledge graphs according to claim 1, characterized in that, The scene wear resistance determination value is obtained, including: According to the humidity holding time, braking interval and disc temperature limit, the bench test is divided into multiple equal braking energy intervals, and the wear amount per unit braking energy and the increase in friction film damage area are calculated after each equal braking energy interval. The intermittent segment in which the wear of unit braking energy increases beyond the weighing error and the increase in the damaged area of ​​the friction film increases beyond the image recognition error is defined as the propagation path backfill segment. The implicit wheel position associated edge weights are corrected based on the degree of deviation between the unit braking energy wear amount and the increase in friction film damage area of ​​the backfill section of the propagation path and the historical wear results, and the scenario wear resistance judgment value is calculated from the corrected implicit wheel position associated edge weights.

10. A knowledge graph-based automotive brake pad wear resistance testing system, used to implement the knowledge graph-based automotive brake pad wear resistance testing method according to any one of claims 1-9, characterized in that, include: The system includes a data acquisition module, a test condition generation module, and a brake pad test output module.

Images