Method and system for identifying internal durability damage of a tire

Abstract

The application provides a tire internal durability damage identification method and system, and belongs to the technical field of data processing, which comprises the following steps: acquiring a dynamic response signal of a tire in an operation process collected by a sensing unit arranged on an inner wall of the tire and constructing a current feature vector; inputting the current feature vector into a reference prediction model to acquire a predicted feature vector; the reference prediction model comprises a mechanism module for representing physical characteristics of the tire and a data-driven module for compensating nonlinear errors; calculating a difference between the current feature vector and the predicted feature vector to generate a residual vector; based on a sliding time window, continuously analyzing statistical features of the residual vector, and when an analysis result meets a preset abnormal condition, determining that the tire is in a durability damage state. The application uses a hybrid model to construct a high-precision normal reference, uses residual time domain consistency to effectively filter out occasional road interference, and realizes accurate identification of internal structure fatigue damage of the tire under the condition of no fault sample.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for identifying internal durability damage in tires. Background Technology

[0002] In the long-term operation of commercial vehicles and heavy-duty vehicles, the internal structure of the tire is subjected to continuous alternating stress, and real-time perception of its health status is crucial to ensuring vehicle operation safety.

[0003] Existing technologies mainly rely on tire pressure monitoring systems (TPMS) to collect macroscopic parameters such as air pressure and temperature inside the tire, or to evaluate tire performance through regular visual inspections and offline destructive tests; some technical solutions attempt to use accelerometers to monitor tire vibration and determine whether there is an abnormality by judging whether the real-time signal amplitude exceeds a set fixed threshold.

[0004] However, the aforementioned existing technologies are insufficient to directly reflect the progressive fatigue damage of internal tire structures such as cord layers or belt layers. Because macroscopic parameters are insensitive to local structural performance degradation, and vibration monitoring methods based on fixed thresholds lack a dynamic description of the tire's normal operating baseline, they cannot distinguish between incidental impacts caused by road bumps and persistent signal distortions caused by damage to the tire's own structure. This leads to false alarms or monitoring delays in complex dynamic conditions, making it impossible to achieve early and accurate identification of internal tire durability damage. Summary of the Invention

[0005] This invention provides a method and system for identifying internal durability damage in tires, which addresses the deficiencies in the prior art and enables early and accurate identification of internal durability damage in tires.

[0006] This invention provides a method for identifying internal durability damage in tires, comprising the following steps: The dynamic response signal of the tire during operation is acquired by the sensing unit arranged on the inner wall of the tire. Feature extraction is performed on the dynamic response signal to construct a current feature vector containing multidimensional features; The current feature vector is input into the benchmark prediction model to obtain the predicted feature vector output by the benchmark prediction model; wherein, the benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors, and the benchmark prediction model is obtained by training the data-driven module with training samples, wherein the training samples are historical feature vectors under normal tire operating conditions; Calculate the difference between the current feature vector and the predicted feature vector to generate a residual vector; Based on a sliding time window, the statistical characteristics of the residual vector are analyzed for continuity, and when the analysis results meet preset abnormal conditions, the tire is determined to be in a state of durability damage.

[0007] According to the present invention, a method for identifying tire internal durability damage includes inputting the current feature vector into a benchmark prediction model to obtain the predicted feature vector output by the benchmark prediction model, comprising: Obtain the current operating parameters of the tires; Using the aforementioned mechanism module, a physical reference response is calculated based on the operating condition parameters and the tire's annular dynamic equation, serving as a reference prediction component; Using the data-driven module, an error correction component for correcting systematic deviations is calculated based on the current feature vector; The baseline prediction component and the error correction component are superimposed to obtain the prediction feature vector.

[0008] According to the present invention, a method for identifying internal tire durability damage is provided, wherein the data-driven module is trained through the following steps: Obtain historical feature vectors of tires under normal operating conditions as training samples; The training samples are input into the mechanism module to obtain the physical reference components output by the mechanism module; Calculate the deviation between the training sample and the physical reference component, and use the deviation as the training target; The network parameters of the data-driven module are iteratively updated with the optimization objective of minimizing the error between the output of the data-driven module and the training target.

[0009] According to a tire internal durability damage identification method provided by the present invention, the step of extracting features from the dynamic response signal and constructing a current feature vector containing multi-dimensional features includes: Calculate the baseline of the mean value of the dynamic response signal over a preset time period; Determine the intersection point between the signal curve of the dynamic response signal and the mean baseline; Calculate the area of ​​the closed region enclosed by the signal curve, the mean baseline, and two adjacent intersection points; Extract the peak and valley values ​​of the dynamic response signal within the preset time period respectively; Calculate the positive deviation of the peak value relative to the mean baseline, and the negative deviation of the valley value relative to the mean baseline; The current feature vector is constructed based on the area of ​​the closed region, the positive deviation, and the negative deviation.

[0010] According to the present invention, a method for identifying internal tire durability damage includes, based on a sliding time window, performing continuous analysis on the statistical characteristics of the residual vector, and determining that the tire is in a state of durability damage when the analysis results meet preset abnormal conditions, comprising: For any moment within the sliding time window, calculate the Euclidean norm of the residual vector to obtain the damage score at the current moment; When the damage score at the current moment exceeds a preset safety threshold, an anomaly count marker is generated; Count the total number of anomaly counters within the sliding time window; When the total number exceeds a preset counting threshold, the tire is determined to be in a state of durability damage.

[0011] According to the present invention, a method for identifying internal tire durability damage includes acquiring dynamic response signals of the tire during operation collected by a sensing unit arranged on the inner wall of the tire, comprising: During tire rolling, the acceleration signal of the inner wall of the tire is received synchronously by all the sensing units; wherein the sensing units are arranged at intervals along the circumferential direction of the inner wall of the tire.

[0012] The present invention also provides a tire internal durability damage identification system, comprising the following modules: The acquisition module is used to acquire the dynamic response signals of the tire during operation collected by the sensing unit arranged on the inner wall of the tire. The feature extraction module is used to extract features from the dynamic response signal and construct a current feature vector containing multi-dimensional features; The prediction module is used to input the current feature vector into the benchmark prediction model and obtain the predicted feature vector output by the benchmark prediction model; wherein, the benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors, and the benchmark prediction model is obtained by training the data-driven module with training samples, wherein the training samples are historical feature vectors under normal tire operating conditions; The residual calculation module is used to calculate the difference between the current feature vector and the predicted feature vector, and generate a residual vector. The discrimination module is used to perform continuous analysis on the statistical characteristics of the residual vector based on a sliding time window, and determine that the tire is in a state of durability damage when the analysis results meet preset abnormal conditions.

[0013] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the tire internal durability damage identification method as described above.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the tire internal durability damage identification method as described above.

[0015] The present invention also provides a computer program product, including a computer program that, when executed by a processor, implements the tire internal durability damage identification method as described above.

[0016] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: By acquiring dynamic response signals collected by sensing units arranged on the inner wall of the tire, the true dynamic behavior of the tire's internal structure under rolling contact can be directly perceived, overcoming the limitation of existing external monitoring technologies that are unable to reflect changes in internal stress. By inputting the extracted multidimensional feature vectors into a hybrid benchmark prediction model containing a mechanism module and a data-driven module, the explicit physical meaning of the physical model and the high-dimensional nonlinear fitting capability of the data-driven model are integrated, constructing a normal state benchmark that accurately matches the current working condition without relying on scarce fault training samples. By calculating the residual vector between the current feature vector and the predicted feature vector, the signal distortion caused by the slight degradation of structural stiffness or damping characteristics is separated from the strong background signal and amplified, achieving a keen capture of early tire durability damage. By performing continuous analysis of the residual statistical characteristics based on the sliding time window, the temporal persistence characteristics of the damage signal are used to effectively filter out occasional interference caused by uneven road surfaces or transient impacts, achieving early and accurate identification of tire internal durability damage. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0018] Figure 1 This is one of the flowcharts of the tire internal durability damage identification method provided by the present invention.

[0019] Figure 2 This is a schematic diagram of the arrangement of sensors on the inner wall of a tire provided by the present invention.

[0020] Figure 3 This is a tire durability test scenario diagram provided by the present invention.

[0021] Figure 4This is a feature sequence data map obtained by the multi-channel sensing unit provided by the present invention.

[0022] Figure 5 This is a schematic diagram of signal feature extraction provided by the present invention.

[0023] Figure 6 This is a schematic diagram illustrating the trend of residual score as a function of sample number based on intelligent tire sensing and hybrid model provided by the present invention.

[0024] Figure 7 This is a schematic diagram of the durability damage alarm results based on a hybrid model and consistency criteria provided in an embodiment of this application.

[0025] Figure 8 This is a schematic diagram of the tire internal durability damage identification system provided by the present invention.

[0026] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0028] It should be noted that in the description of this invention, the terms "comprising," "including," or any other variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element. The terms "upper," "lower," etc., indicating orientation or positional relationships according to the accompanying drawings, are only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the system or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0029] The terms "first," "second," etc., used in this invention are used to distinguish similar objects, not to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that embodiments of the invention can be implemented in orders other than those illustrated or described herein, and the objects distinguished by "first," "second," etc., are generally of the same class, without limiting the number of objects; for example, a first object can be one or more. Furthermore, "and / or" indicates at least one of the connected objects, and the character " / " generally indicates that the preceding and following objects are in an "or" relationship.

[0030] The following is combined Figures 1 to 9 This invention describes the tire internal durability damage identification method, system, electronic device, storage medium, and computer program product provided by the present invention.

[0031] This application provides a method for identifying internal tire durability damage. This specification describes the application of this method to a tire internal durability damage identification system (hereinafter referred to as the system) as an example. The tire internal durability damage identification system can be configured in the vehicle's electronic control unit, or it can be configured in a remote server or a dedicated tire monitoring terminal.

[0032] Reference Figure 1 , Figure 1 This is one of the flowcharts illustrating the tire internal durability damage identification method provided by this invention. For example... Figure 1 As shown, the method for identifying internal tire durability damage specifically includes steps 101 to 105: Step 101: Acquire the dynamic response signal of the tire during operation collected by the sensing unit arranged on the inner wall of the tire; Step 102: Extract features from the dynamic response signal and construct a current feature vector containing multidimensional features; Step 103: Input the current feature vector into the benchmark prediction model and obtain the predicted feature vector output by the benchmark prediction model; wherein, the benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors. The benchmark prediction model is obtained by training the data-driven module with training samples, which are historical feature vectors under normal tire operating conditions. Step 104: Calculate the difference between the current feature vector and the predicted feature vector, and generate the residual vector; Step 105: Based on the sliding time window, perform continuity analysis on the statistical characteristics of the residual vector, and determine that the tire is in a state of durability damage when the analysis results meet the preset abnormal conditions.

[0033] The system first acquires the dynamic response signals of the tire during operation, collected by the sensing units arranged on the inner wall of the tire.

[0034] The sensing unit is arranged circumferentially along the inner wall of the tire, enabling it to sense changes in acceleration or vibration of the tire's inner wall in real time as the tire rolls with the vehicle. The dynamic response signal reflects the tire's transient response characteristics under different speeds, loads, and road surface excitations.

[0035] Furthermore, the system extracts features from the dynamic response signal and constructs a current feature vector containing multidimensional features.

[0036] Multidimensional features are used to quantitatively describe the waveform characteristics of dynamic response signals in the time or frequency domain. The system extracts key indicators that characterize the tire's structural state through numerical analysis of the dynamic response signal, and combines these key indicators to form the current feature vector.

[0037] Furthermore, the system inputs the current feature vector into the baseline prediction model to obtain the predicted feature vector output by the baseline prediction model.

[0038] The baseline prediction model is a hybrid architecture model, comprising a mechanistic module and a data-driven module. The mechanistic module characterizes tire physical properties based on physical laws and tire structural parameters, providing a theoretical physical benchmark. The data-driven module learns nonlinear mapping relationships to compensate for nonlinear errors.

[0039] The baseline prediction model is trained on the data-driven module using training samples from tires operating normally. When the tire is in normal operating condition, the system uses collected historical normal data to train the baseline prediction model, enabling it to learn the inherent mapping relationships that the tire's various features should satisfy under healthy conditions.

[0040] Furthermore, the system calculates the difference between the current feature vector and the predicted feature vector to generate a residual vector.

[0041] The residual vector reflects the degree to which the tire's current actual operating condition deviates from the normal state predicted by the baseline prediction model. Under ideal normal conditions, the value of the residual vector should be close to zero or within a very small fluctuation range.

[0042] Finally, based on a sliding time window, the system performs a continuity analysis on the statistical characteristics of the residual vector, and determines that the tire is in a state of durability damage when the analysis results meet the preset abnormal conditions.

[0043] The system sets a sliding time window of fixed length and statistically observes the residual vectors at multiple consecutive time points within this window. The system distinguishes between incidental road disturbances and genuine structural damage by determining whether the statistical characteristics of the residual vectors consistently exhibit an abnormal pattern over a period of time. When the analysis results indicate that the residual vectors continuously meet preset abnormal conditions within the sliding time window, the system determines that irreversible durability damage has occurred to the tire's internal structure.

[0044] The tire internal durability damage identification method provided in this embodiment establishes a high-precision tire normal state benchmark by constructing a benchmark prediction model that includes a mechanism module and a data-driven module. The mechanism module ensures physical interpretability, while the data-driven module compensates for nonlinear errors under complex working conditions. By calculating the residual vector between the current feature vector and the predicted feature vector and using the residual vector as the basis for damage identification, unsupervised anomaly detection without requiring a large number of fault samples is achieved. Furthermore, by performing continuity analysis on the residual vector based on a sliding time window, instantaneous signal fluctuations caused by external factors such as road bumps can be effectively filtered out, significantly reducing the false alarm rate and achieving stable and reliable identification of tire internal durability damage.

[0045] In one embodiment of this application, the steps for acquiring dynamic response signals are described in detail.

[0046] During tire rolling, the system receives acceleration signals from the inner wall of the tire, which are synchronously collected by all sensing units; the sensing units are arranged at intervals along the circumference of the inner wall of the tire.

[0047] Specifically, the system utilizes multiple microelectromechanical system (MEMS) accelerometers mounted on the surface of the tire's inner liner as sensing units. These sensing units are evenly or non-uniformly spaced along the circumference of the tire's inner wall.

[0048] In a preferred embodiment, refer to Figure 2 , Figure 2 This is a schematic diagram of the tire inner wall sensor arrangement provided by the present invention. The preferred number of sensors is four, arranged at intervals along the circumference of the tire inner wall to form a distribution structure covering the entire tire circumference. Each sensor is fixedly installed on the surface of the tire inner wall rubber layer and rotates synchronously with the tire, enabling real-time sensing of the dynamic response information generated by the tire during rolling, loading, and deformation.

[0049] By employing the aforementioned circumferential distribution method, multiple sensors can acquire dynamic response signals from the tire's interior from different angles and positions, thereby avoiding information distortion caused by a single sensor due to local structural differences or occasional interference. During durable operation, tire internal structural damage often exhibits localized and asymmetric characteristics. The circumferential arrangement of multiple sensors can effectively capture response differences at different locations, improving the ability to perceive abnormal conditions within the tire.

[0050] Furthermore, the dynamic response signals collected by each sensor can be individually extracted and analyzed, or fused to construct an overall feature distribution model of the tire under normal operating conditions. When fatigue damage occurs in the internal structure of the tire during durable operation, the response characteristics of sensors at different locations will exhibit consistent or locally enhanced abnormal changes. Through joint analysis of multi-sensor data, the internal durability damage state of the tire can be identified more accurately.

[0051] The above-mentioned sensor arrangement is simple in structure, reliable in installation, does not affect the normal performance of the tire, and can achieve continuous monitoring of internal tire durability damage without changing the external structure of the tire. It is suitable for smart tires and tire durability testing scenarios.

[0052] Reference Figure 3 , Figure 3 This is a diagram of a tire durability test scenario provided by the present invention. The tire durability test is conducted on the MTS tire bench durability testing system. The test bench applies stable and controllable rotational speed, load, and running time to the tire under test through a loading system, drive system, and control system to simulate the long-term durability conditions of the tire during actual vehicle operation. During the test, the tire is mounted on the main shaft of the test bench and forms stable contact with the loading roller of the test bench. The continuous rolling of the tire is achieved by the rotation of the roller.

[0053] Before the test, the tire under test was installed, positioned, and dynamically balanced. A smart tire sensor was attached to a predetermined position on the inner wall of the tire, enabling it to rotate synchronously with the tire and collect internal dynamic response signals. The sensor system was then initialized and calibrated to ensure the stability and consistency of the data acquisition process. The tire was inflated to a predetermined pressure, and a corresponding vertical load was applied under the action of the loading system, placing the tire under the target durability condition.

[0054] During the test, the test bench continuously drove the tire to run according to the set speed and load conditions, and the control system maintained the stability of the speed and load. During long-term rolling, the tire's internal structure is constantly subjected to cyclic alternating stress, gradually leading to fatigue damage. Sensors continuously collected the tire's inner wall acceleration signals throughout the durability test and transmitted the collected data in real time to the data acquisition system for storage.

[0055] As the test duration increases, the test bench can adjust the tire load or extend the operating time step by step according to the test plan to accelerate the fatigue evolution process of the tire's internal structure. Throughout the test, the test bench monitors the tire's operating status in real time, and can stop the test immediately if any abnormal conditions or safety risks occur. The test continues until the tire enters a stage of obvious durability damage or failure, forming a complete durability data sequence covering the normal operation and failure stages of the tire.

[0056] Through the above experimental process, continuous operating data of the tire under controllable durability conditions were obtained, providing a reliable experimental data foundation for subsequent feature extraction based on intelligent tire sensing, hybrid model modeling, and identification of internal tire durability damage.

[0057] Reference Figure 4 , Figure 4 This is a feature sequence data map obtained by the multi-channel sensing unit provided by the present invention. Figure 4 The diagram shows four types of characteristic sequence data collected and calculated by a multi-channel intelligent sensing unit deployed inside the tire during the tire durability test failure phase.

[0058] In the figure, the horizontal axis of each data set represents the sample number, corresponding to the continuous sampling time of the tire during durability operation. The vertical axis of each data set represents the change in the numerical amplitude of each characteristic quantity. The above data covers the complete process of the tire gradually developing from a relatively stable operating state to a stage of durability damage, and can reflect the characteristics of the tire's internal structural response evolving over time.

[0059] The first set of data reflects the average deviation of the tire's internal dynamic response signal from the mean within a local time window. The overall trend is relatively stable, but there are abrupt amplitude changes and increased fluctuations in local sections, indicating that the tire's internal structure gradually produces an unstable response in the later stages of durability.

[0060] The second set of data reflects the deviation characteristics of the mean dynamic response in another direction or position. It fluctuates less in the early stage of durability operation, but a large number of spike responses appear in the later stage, which reflects the characteristics of increased local impact or structural discontinuity inside the tire.

[0061] The third set of data also shows fluctuations around the zero mean, but the negative deviation gradually increases with durability time, indicating that the tire's response to dynamic excitation is weakened after the internal structural stiffness and damping characteristics change.

[0062] The fourth set of data is an area-type feature constructed based on the intersection of the signal and the mean. During the durability operation, it shows a trend of changing from small fluctuations to large fluctuations, reflecting the increasing cumulative deviation of the tire's internal dynamic response from the baseline state.

[0063] In summary, the four types of features exhibit significant differences in both time and amplitude scales, but they share a consistent evolutionary trend during the durability failure stage: a gradual transition from a relatively stable state to a highly volatile and deviating state. This original feature set provides fundamental data support for subsequent construction of tire normal state mapping relationships, calculation of feature deviations, and durability damage identification, effectively reflecting the damage evolution process of tire internal structures under long-term durability operating conditions.

[0064] Furthermore, the sensing unit can directly measure the local acceleration of the tire's inner wall in a rotating coordinate system. The acceleration signal acquired by the sensing unit includes the relative acceleration component caused by tire structural vibration, the centripetal acceleration component caused by rotational motion, and the Coriolis acceleration component. Let the position vector of the sensing unit in a fixed coordinate system rotating with the tire be... The tire angular velocity is The localized minute vibration displacement of the tire inner wall is The acceleration of the sensing unit in the inertial frame is then... This can be represented as the superposition of rotational rigid body motion and relative motion, that is: ; in, For the centripetal term; for The first derivative of is the Coriolis term; for The second derivative of represents the relative acceleration generated by structural vibration, i.e., the dynamic response caused by the deformation of the internal structure of the tire.

[0065] For scenarios involving constant-speed rolling where the vibration displacement is much smaller than the magnitude of the position vector, the vibration displacement in the centripetal term of the acceleration signal... If it can be treated as a small quantity, then the approximate observation model can be expressed as: ; The acceleration signal received by the system is a comprehensive representation of the aforementioned physical quantities. Because the sensors are mounted on the inner wall and distributed at different circumferential positions (4 points), the system can simultaneously or time-division multiplex the dynamic response of the tire's inner wall from different phase angles, thus avoiding monitoring blind spots caused by the location limitations of a single sensor. The first... The observations from each sensor are written as: in Indicates the first Sensor mounting orientation of each sensor (projecting three-dimensional acceleration onto the sensor measurement axis). For the first The measurement noise and error terms of each sensor, such as pasting / installation disturbances.

[0066] The technical solution provided in this embodiment, by arranging multiple sensing units at intervals along the circumference of the tire's inner wall and simultaneously acquiring acceleration signals, can capture the dynamic response of the tire during rolling in all directions. Utilizing the high sensitivity of acceleration signals to structural vibration, combined with the spatial coverage advantage of multi-point distribution, this solution can effectively capture local stiffness changes or damping characteristic alterations caused by fatigue damage to the tire's internal structure, providing a rich and high signal-to-noise ratio raw data foundation for subsequent feature extraction and damage identification.

[0067] In one embodiment of this application, the specific steps for feature extraction are described in detail. Step 102 specifically includes the following steps: Step 201: Calculate the mean baseline of the dynamic response signal over a preset time period; Step 202: Determine the intersection point between the signal curve of the dynamic response signal and the mean baseline; Step 203: Calculate the area of ​​the closed region enclosed by the signal curve, the mean baseline, and two adjacent intersection points; Step 204: Extract the peak and valley values ​​of the dynamic response signal within a preset time period; Step 205: Calculate the positive deviation of the peak value from the mean baseline and the negative deviation of the trough value from the mean baseline; Step 206: Construct the current feature vector based on the area of ​​the closed region, the positive deviation, and the negative deviation.

[0068] The system first calculates the baseline of the mean value of the dynamic response signal over a preset time period. The preset time period can be a sliding window. The sliding window contains N There are sampling points, and the time series set corresponding to this window is . .

[0069] The system processes the acceleration signal sequence within this window. Calculate the arithmetic mean to obtain the window mean. .

[0070] Reference Figure 5 , Figure 5 This is a schematic diagram of signal feature extraction provided by the present invention. For example... Figure 5 The schematic diagram of signal feature extraction illustrates the geometric meaning of this process. The gray curve in the diagram represents the acquired dynamic response signal. The yellow dashed line crossing the center of the signal fluctuation in the diagram represents the calculated mean baseline, whose value corresponds to the window mean. This is used to reflect the local DC component or quasi-static reference level of the signal within the current time period. The formula for calculating the window mean is: ; The system determines the intersection point between the signal curve of the dynamic response signal and the mean baseline. For example... Figure 5 As shown, the signal curve crosses the mean baseline multiple times over time. The system identifies, through numerical analysis, that the signal curve value equals... At all times, the intersection point of the signal curve and the mean baseline is determined.

[0071] To quantify the local fluctuation characteristics of the signal, the system identifies four sets of key geometric features in the figure, corresponding to the first, second, third, and fourth columns, respectively.

[0072] The system extracts the peak and valley values ​​of the dynamic response signal within a preset time period, and calculates the positive deviation of the peak value from the mean baseline and the negative deviation of the valley value from the mean baseline.

[0073] See details Figure 5 , Figure 5 The feature marked "first column" represents the vertical distance between a local peak located on the left side of the time window and the mean baseline shown by the yellow dashed line. This distance characterizes the positive instantaneous impact strength of the signal relative to the reference level at that moment.

[0074] Figure 5 The feature marked "second column" represents the vertical distance between the global minimum (i.e., valley) within the time window and the mean baseline shown by the yellow dashed line. This distance characterizes the maximum negative deviation of the signal relative to the baseline level.

[0075] Figure 5 The feature marked "third column" indicates the vertical distance between another local peak located to the right of the time window and the mean baseline shown by the yellow dashed line. This distance characterizes the positive instantaneous impact intensity during the subsequent rebound process.

[0076] The system selects the maximum value of the positive height difference shown in the first and third columns above as the positive deviation. The negative height difference shown in the second column is selected as the negative deviation. .

[0077] Specifically, the system extracts the peak and trough values ​​of the dynamic response signal within a preset time period. The system operates within a sliding window. Internal search signal s ( t Maximum value in the positive direction And the minimum value in the negative direction . .

[0078] The system calculates the positive deviation of the peak value from the mean baseline, and the negative deviation of the trough value from the mean baseline. The specific calculation formula is as follows: .

[0079] The system determines the intersection point between the signal curve of the dynamic response signal and the mean baseline, and calculates the area of ​​the closed region enclosed by the signal curve, the mean baseline, and two adjacent intersection points. Figure 5 The feature labeled "Fourth Column" visually illustrates the geometric definition of this area-type feature. The blue shaded area indicated by the fourth column is a closed geometric shape formed by the concave signal curve segment, the mean baseline indicated by the yellow dashed line above, and the two intersection points of the signal curve and the mean baseline on the left and right sides. The area of ​​this closed region intuitively reflects the cumulative deviation or fluctuation energy of the signal relative to the mean within that local time period. (System definition window: Deviation) The area of ​​the closed region is obtained by integrating or summing the absolute value of the deviation over the window time. The formula for calculating the area of ​​a closed region is: ; in, The sampling time interval is defined as . This area-type feature comprehensively reflects the amplitude and duration information of the signal waveform and is highly sensitive to waveform distortion.

[0080] The system is based on the area of ​​the enclosed region. Positive deviation and negative deviation The system constructs the current feature vector. Specifically, it combines the magnitude deviation features represented by the first, second, and third columns with the area energy features represented by the fourth column to form a current feature vector containing multi-dimensional features. .

[0081] The technical solution provided in this embodiment constructs a multi-dimensional feature vector containing area features (fourth column), positive deviation (first and third columns), and negative deviation (second column) by combining the geometric feature extraction method shown in the accompanying drawings. The amplitude features in the first to third columns can sensitively capture the instantaneous impact enhancement caused by the decrease in stiffness of the tire's internal structure; the area features in the fourth column can effectively quantify the cumulative effect of waveform distortion. This feature extraction method based on geometric morphology can comprehensively characterize the abnormal changes in the tire's dynamic response signal from both the amplitude and time domains, significantly improving the sensitivity of identifying early durability damage inside the tire.

[0082] In one embodiment of this application, the steps for obtaining a predicted feature vector using a benchmark prediction model are described in detail. Specifically, the steps include the following: Step 301: Obtain the current operating parameters of the tire; Step 302: Using the mechanism module, calculate the physical reference response based on the operating condition parameters and the tire's ring dynamic equation, and use it as the reference prediction component; Step 303: Using the data-driven module, calculate the error correction component for correcting systematic biases based on the current feature vector; Step 304: Superimpose the baseline prediction component and the error correction component to obtain the prediction feature vector.

[0083] The system acquires the current operating parameters of the tires. These parameters include tire speed, inflation pressure, and load information. These parameters can be obtained from the vehicle bus or calculated by analyzing sensor signals.

[0084] The system utilizes a mechanism module to calculate the physical baseline response based on operating condition parameters and the tire's annular dynamics equations, serving as the baseline prediction component. The mechanism module includes a simplified tire annular dynamics model to describe the radial displacement field in the tire's circumferential direction. Let... θ Circumferential angular coordinates For the radial displacement of the ring belt, the dynamic equation of the ring belt can be expressed as: ; in, It is the equivalent circumferential mass per unit ring. For equivalent damping, This represents the radial "foundation" stiffness term formed by the combined effects of air pressure and the tire body. This indicates circumferential tension / prestress related terms. This represents the bending stiffness term. (Right end) This indicates the equivalent effect of the tire's internal air pressure on the tire ring. This represents the equivalent contact force density formed by the contact patch constraint and the road surface excitation. The mechanism module determines the angular frequency based on current operating parameters (e.g., angular frequency determined by rotational speed, air pressure, etc.). Solving the above equations or their simplified forms yields the theoretical displacement or acceleration response characteristics, i.e., the physical reference response, denoted as . This physical reference response reflects the dynamic characteristics that a tire should exhibit under ideal linear and uniform assumptions.

[0085] The system utilizes a data-driven module to calculate error correction components based on the current feature vector to correct systematic biases. This data-driven module employs a neural network structure to capture nonlinear factors that mechanistic models cannot cover, such as the nonlinear viscoelasticity of rubber materials, contact friction, and sensor installation errors. Let the input feature vector at the current moment be... The data-driven module uses mapping functions The error correction component was calculated. : in, This refers to the set of model parameters learned by the data-driven module during the training phase. Based on the current input feature vector, the data-driven module predicts the deviation of the mechanism module's output from the true normal state.

[0086] The system superimposes the baseline prediction component and the error correction component to obtain the predicted feature vector. Specifically, the system uses the physical baseline response... With error correction components Add them together to get the final mixed prediction output. : ; Predicting feature vectors This represents the theoretical value that the feature vector should possess under the current working conditions, assuming the tire is in a normal and healthy state.

[0087] The technical solution provided in this embodiment adopts a hybrid modeling strategy of "mechanism framework + data-driven correction". By introducing the ring dynamic equation through the mechanism module, a baseline for the model's physical interpretability and generalization ability under a wide range of operating conditions is ensured. The data-driven module learns and compensates for model residuals, effectively correcting systematic biases caused by idealized assumptions. This hybrid prediction approach enables the baseline prediction model to possess both the robustness of a physical model and the high accuracy of a data-driven model, enabling more accurate reconstruction of the normal state characteristics of tires under complex and varying operating conditions, laying a solid foundation for subsequent high-sensitivity damage identification.

[0088] In one embodiment of this application, the training steps of the data-driven module are described in detail. Specifically, they include the following steps: Step 401: Obtain the historical feature vector of the tire under normal operating conditions as training samples; Step 402: Input the training samples into the mechanism module to obtain the physical reference components output by the mechanism module; Step 403: Calculate the deviation between the training samples and the physical reference components, and use the deviation as the training target; Step 404: With minimizing the error between the output of the data-driven module and the training target as the optimization objective, iteratively update the network parameters of the data-driven module.

[0089] During the training phase, the system first acquires historical feature vectors of the tires under normal operating conditions as training samples. Under normal operating conditions (i.e., before fatigue damage or structural failure), the tires operate under different loads, air pressures, and speeds, and sensors collect a large amount of data and extract feature vector sequences. This data from healthy tires constitutes the positive sample set.

[0090] The system inputs training samples into the mechanism module to obtain the physical reference components output by the mechanism module. For each time step... k Input feature vector The mechanism module calculates the theoretical physical prediction value, i.e. the physical reference component, based on the corresponding operating parameters and using the aforementioned annular dynamic equation. This component represents the characteristic value that a tire should exhibit under the assumptions of an ideal physical model.

[0091] The system calculates the deviation between the training samples and the physical reference components, and uses this deviation as the training objective. Specifically, the system calculates the normal feature vectors actually collected. (or its corresponding target output dimension) and physical reference components The difference between these values ​​reflects systematic errors that the mechanistic model fails to explain, such as material nonlinearity or installation errors. This deviation is set as the target value that the data-driven module should learn to fit.

[0092] The system optimizes by minimizing the error between the output of the data-driven module and the training objective, iteratively updating the network parameters of the data-driven module. The output of the data-driven module (e.g., a neural network) is... The system defines the loss function as the mean square error between the correction predicted by the data-driven module and the actual deviation calculated above. The following objective function is minimized through an optimization algorithm: ,in, Let W be the number of training samples, and W be the network parameters to be optimized. Through backpropagation or other optimization strategies, the system continuously adjusts the parameters W, enabling the data-driven module to accurately predict the residuals of the mechanistic model.

[0093] The technical solution provided in this embodiment employs a supervised learning strategy based on physical bias to train the data-driven module. Unlike directly learning the input-output mapping, this method focuses on learning the "parts that the physical model cannot explain," i.e., systematic biases. This training method significantly reduces the learning difficulty of the data-driven module, enabling it to converge with fewer normal samples. Furthermore, after training, the hybrid model can accurately describe the complex nonlinear behavior of tires under normal conditions, providing a high-confidence reference benchmark for anomaly detection during online monitoring.

[0094] In one embodiment of this application, the continuity analysis and damage determination steps based on a sliding time window are described in detail. Step 105 specifically includes the following steps: Step 501: For any moment in the sliding time window, calculate the Euclidean norm of the residual vector to obtain the damage score at the current moment; Step 502: When the damage score at the current moment exceeds the preset safety threshold, generate an anomaly count marker; Step 503: Count the total number of anomaly markers within the sliding time window; Step 504: When the total number exceeds the preset counting threshold, the tire is determined to be in a state of durability damage.

[0095] The system calculates the Euclidean norm of the residual vector at any point within the sliding time window to obtain the damage score for that moment. During subsequent tire durability testing, the system calculates the current feature vector in real time. With predicted feature vectors The residual vector between , To quantify the current degree of deviation, the system calculates the Euclidean norm (L2 norm) of the residual vector as a standardized damage score. : Damage score It comprehensively reflects the overall deviation of the tire's current operating state from the normal baseline state.

[0096] The system generates an anomaly count marker when the damage score at the current moment exceeds a preset safety threshold. The preset safety threshold can be determined through statistical analysis of the damage scores of normal training samples (e.g., using the 99th percentile). Let the anomaly indicator variable be... When the first j When the damage score at any sampling time exceeds the preset safety threshold, The value is 1 (indicating an abnormality), otherwise it is 0 (indicating normality).

[0097] The system counts the total number of anomaly markers within the sliding time window. Let the length of the sliding time window be...H That is, including H The system sums the anomaly indicator variables within the window to calculate the total number of anomalies within the window. : The total This reflects the frequency of abnormal deviations in tire condition over a recent period.

[0098] The system determines that a tire is in a state of durability damage when the total number of abnormal points exceeds a preset counting threshold. The system sets a minimum threshold for the number of abnormal points. As a counting threshold. The total number of outliers within the sliding window. When the system determines that the current anomaly is not an occasional event caused by random road surface disturbance, but rather has persistent structural damage characteristics, it outputs a durability damage assessment result. : in This is an indicator function used to map the anomaly count results to a binary judgment output. Once a durability damage state is determined, the system can trigger an alarm or log the event.

[0099] The technical solution provided in this embodiment introduces a consistency discrimination mechanism based on a sliding time window. Unlike single-point threshold alarms, this scheme requires the abnormal state to have a certain duration over time before triggering the final judgment. It utilizes the irreversible and continuously evolving physical characteristics of tire fatigue damage, while road bumps and other disturbances usually manifest as transient impacts. Thus, while ensuring high sensitivity to real damage, it greatly suppresses false alarms and significantly improves the robustness and reliability of the recognition results in complex dynamic environments.

[0100] This application provides a detailed description of the residual score change trend after processing based on a hybrid model. (Refer to...) Figure 6 , Figure 6 This is a schematic diagram illustrating the trend of residual score as a function of sample number based on intelligent tire sensing and hybrid model provided by the present invention.

[0101] Figure 6 The horizontal axis is labeled "Sample index (Normal then Fail)," representing the sample number of consecutive sampling, covering the entire process of the tire gradually transitioning from normal durability operation (left area of ​​the figure) to the failure stage (right area of ​​the figure). The vertical axis is labeled "Residual score (4-dim norm)," representing the comprehensive residual score calculated from the multi-dimensional feature vector. The magnitude of this score reflects the degree of deviation of the current tire operating state from the pre-built normal state mapping model.

[0102] The graph contains multiple indicator curves and boundary markers. The light gray thin line corresponds to "Rawscore" in the legend, representing the raw residual score calculated at each sampling time. The thick red solid line corresponds to "Slidingmean" in the legend, representing the trend curve after applying a sliding average to the raw residual score, with the sliding window width set to 500 sample points. The black horizontal dashed line corresponds to "Threshold" in the legend, representing the preset safety threshold. The vertical dashed line corresponds to "Boundary" in the legend and is labeled "Normal / Fail boundary," indicating the physical boundary when the tire transitions from a normal state to a failure state.

[0103] During normal tire operation (i.e., the area to the left of the vertical dashed line), the residual score calculated by the system is generally at a low level. Although occasional instantaneous spikes appear in the original residual score curve due to random factors such as road surface unevenness and local impacts from tire rolling, its moving average curve remains stable and significantly lower than the preset safety threshold. This statistical characteristic indicates that when there is no obvious fatigue damage to the internal tire structure, the benchmark prediction model trained based on normal state training samples can accurately reconstruct or predict the current output characteristics. The residual between the model output and the actual observation value mainly comes from unavoidable random disturbances and operating condition fluctuations, rather than systematic structural deviations.

[0104] As the endurance test continued, as the tire gradually crossed the boundary between normal and failure stages and entered the failure phase (i.e., the area to the right of the vertical dotted line), the statistical characteristics of the residual score changed significantly. The peak amplitude of the original residual score increased dramatically, and the frequency of high-amplitude peaks became more concentrated, indicating a continuous upward trend in the overall baseline level of the residual score. Correspondingly, the moving average curve, representing the trend change, gradually approached and eventually exceeded the preset safety threshold during the failure stage. This phenomenon indicates that with the accumulation of fatigue damage, the physical parameters of the tire's internal structures, such as cords and belt layers, changed, disrupting the normal mapping relationship between characteristic quantities and thus generating a non-negligible systematic prediction error.

[0105] Furthermore, in the transition region between normal and failure states, the residual score exhibits an evolutionary characteristic shifting from "low mean, occasional spikes" to "high mean, continuous fluctuations." This characteristic objectively reflects the transition process of tire durability damage from its early initiation to its rapid expansion stage. The system introduces a consistency criterion based on a sliding time window, utilizing the moving average curve or outlier count to mitigate the risk of false alarms caused by occasional road impacts during the normal stage, while maintaining a high sensitivity response to persistent abnormal states during the failure stage.

[0106] The technical solution provided in this embodiment, through long-term continuous monitoring and trend analysis of residual scores, can keenly detect statistical characteristic abrupt changes caused by internal structural performance degradation before obvious abnormalities such as bulges and cracks appear on the tire's appearance. This method, based on a combination of hybrid model residual and time consistency analysis, effectively distinguishes between environmental disturbances and structural damage, providing reliable data support for the early identification and safety warning of tire durability damage.

[0107] Reference Figure 7 , Figure 7 This is a schematic diagram of durability damage alarm results based on a hybrid model and consistency criteria provided in an embodiment of this application. Combined with... Figure 7 This application provides a detailed description of the alarm results generated by the system during actual testing.

[0108] Figure 7 The horizontal axis is labeled "Sample index (Normal then Fail)," representing the sequential number of consecutive samples arranged in chronological order; the vertical axis is labeled "Residual score (4-dim norm)," representing the comprehensive score index calculated based on the multidimensional feature residuals.

[0109] The solid black line in the diagram corresponds to "Score" in the legend, representing the damage score curve calculated in real time; the horizontal black dashed line corresponds to "Threshold" in the legend, representing the system's preset anomaly detection threshold (the example value in the diagram is 0.863); the vertical gray dashed line corresponds to "Boundary" in the legend, representing the physical boundary between the tire's normal state and its failure state; the solid red line in the diagram corresponds to "Alarm(red)" in the legend, representing the signal segment that the system determines as durability damage and triggers an alarm. Furthermore, "sum_30>=10" in the diagram title indicates the specific parameter setting for the consistency criterion, meaning that within a sliding window of length 30, if the number of points exceeding the threshold is greater than or equal to 10, an alarm is triggered.

[0110] During normal tire operation (to the left of the vertical dashed line), although several high-amplitude instantaneous peaks (i.e., the scattered black lines exceeding the threshold in the figure) appeared in the damage scoring curve due to external stimuli such as uneven road surface and tire bumps, under the constraint of the time consistency criterion, these discrete points exceeding the threshold did not form a sufficient density within the sliding window. Therefore, the system did not generate a continuous red alarm signal. This indicates that the method provided in this embodiment can effectively filter out false anomalies caused by occasional operating condition disturbances, with an extremely low false alarm rate.

[0111] As the durability test continued, just before the tire entered the failure stage (near the vertical dotted line and the area to the right), the statistical characteristics of the comprehensive scoring index changed significantly. The fluctuation amplitude of the damage score increased dramatically, and the frequency of high-amplitude scores increased significantly, causing the number of sampling points exceeding the preset threshold to meet the preset abnormal conditions (e.g., sum>=10) within a continuous time window. At this point, the system determined that irreversible structural damage had occurred inside the tire and began to output continuous alarm signals, i.e., the dense red area in the figure.

[0112] It is worth noting that the alarm segment begins before catastrophic tire failure or any visible abnormalities occur. This means the system successfully detected early signs of tire internal structural performance degradation, providing early warning of durability damage. The alarm signal is not triggered by a single anomaly, but rather consists of a continuous sequence of samples, objectively reflecting that tire internal structural performance degradation is a continuous evolutionary process, not a random event.

[0113] The technical solution provided in this embodiment constructs a robust alarm mechanism by combining a residual-based scoring index with a sliding window-based time consistency criterion. This mechanism not only accurately distinguishes between occasional operational disturbances and actual durability damage, ensuring high monitoring sensitivity, but also significantly improves the stability and reliability of alarm results. By outputting clear and continuous alarm information before tire failure occurs, the system provides timely and effective evidence for tire condition monitoring, maintenance decisions, and vehicle operation safety management.

[0114] Reference Figure 8 , Figure 8 This is a schematic diagram of the system provided by the present invention. The system includes: The acquisition module is used to acquire the dynamic response signals of the tire during operation collected by the sensing unit arranged on the inner wall of the tire. The feature extraction module is used to extract features from the dynamic response signal and construct a current feature vector containing multi-dimensional features. The prediction module is used to input the current feature vector into the benchmark prediction model and obtain the predicted feature vector output by the benchmark prediction model. The benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors. The benchmark prediction model is obtained by training the data-driven module with training samples, which are historical feature vectors under normal tire operating conditions. The residual calculation module is used to calculate the difference between the current feature vector and the predicted feature vector, and generate a residual vector. The discrimination module is used to perform continuity analysis on the statistical characteristics of the residual vector based on the sliding time window, and to determine that the tire is in a state of durability damage when the analysis results meet the preset abnormal conditions.

[0115] In one possible implementation, the system further includes a training module; the training module is used for: Obtain historical feature vectors of tires under normal operating conditions as training samples; Input the training samples into the mechanism module to obtain the physical reference components output by the mechanism module; Calculate the deviation between the training samples and the physical reference components, and use the deviation as the training target; The network parameters of the data-driven module are iteratively updated with the goal of minimizing the error between the output of the data-driven module and the training objective.

[0116] It should be noted that the tire internal durability damage identification system provided by the present invention can execute the tire internal durability damage identification method of any of the above embodiments during specific operation, which will not be elaborated in this embodiment.

[0117] Figure 9 This is a schematic diagram of the structure of the electronic device provided by the present invention, such as... Figure 9 As shown, the electronic device may include a processor 910, a communication interface 920, a memory 930, and a communication bus 940, wherein the processor 910, the communication interface 920, and the memory 930 communicate with each other via the communication bus 940. The processor 910 can call logical instructions in the memory 930 to execute the tire internal durability damage identification method provided in the above embodiments.

[0118] Furthermore, the logical instructions in the aforementioned memory 930 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0119] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions, and when the program instructions are executed by a computer, the computer is able to execute the tire internal durability damage identification method provided in the above embodiments.

[0120] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to perform the tire internal durability damage identification method provided in the above embodiments.

[0121] The system embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0122] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, 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 computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., including several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods of various embodiments or some parts of embodiments.

[0123] 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 of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for identifying internal durability damage in a tire, characterized in that, include: The dynamic response signal of the tire during operation is acquired by the sensing unit arranged on the inner wall of the tire. Feature extraction is performed on the dynamic response signal to construct a current feature vector containing multidimensional features; The current feature vector is input into the benchmark prediction model to obtain the predicted feature vector output by the benchmark prediction model; wherein, the benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors, and the benchmark prediction model is obtained by training the data-driven module with training samples, wherein the training samples are historical feature vectors under normal tire operating conditions; Calculate the difference between the current feature vector and the predicted feature vector to generate a residual vector; Based on the sliding time window, the statistical characteristics of the residual vector are analyzed for continuity, and when the analysis results meet the preset abnormal conditions, the tire is determined to be in a state of durability damage.

2. The tire internal durability damage identification method according to claim 1, characterized in that, The step of inputting the current feature vector into the benchmark prediction model and obtaining the predicted feature vector output by the benchmark prediction model includes: Obtain the current operating parameters of the tires; Using the aforementioned mechanism module, a physical reference response is calculated based on the operating condition parameters and the tire's annular dynamic equation, serving as a reference prediction component; Using the data-driven module, an error correction component for correcting systematic deviations is calculated based on the current feature vector; The baseline prediction component and the error correction component are superimposed to obtain the prediction feature vector.

3. The tire internal durability damage identification method according to claim 2, characterized in that, The data-driven module is trained through the following steps: Obtain historical feature vectors of tires under normal operating conditions as training samples; The training samples are input into the mechanism module to obtain the physical reference components output by the mechanism module; Calculate the deviation between the training sample and the physical reference component, and use the deviation as the training target; The network parameters of the data-driven module are iteratively updated with the optimization objective of minimizing the error between the output of the data-driven module and the training target.

4. The tire internal durability damage identification method according to claim 1, characterized in that, The step of extracting features from the dynamic response signal and constructing a current feature vector containing multidimensional features includes: Calculate the baseline of the mean value of the dynamic response signal over a preset time period; Determine the intersection point between the signal curve of the dynamic response signal and the mean baseline; Calculate the area of ​​the closed region enclosed by the signal curve, the mean baseline, and two adjacent intersection points; Extract the peak and valley values ​​of the dynamic response signal within the preset time period respectively; Calculate the positive deviation of the peak value relative to the mean baseline, and the negative deviation of the valley value relative to the mean baseline; The current feature vector is constructed based on the area of ​​the closed region, the positive deviation, and the negative deviation.

5. The method for identifying internal tire durability damage according to claim 1, characterized in that, The step of performing continuous analysis on the statistical characteristics of the residual vector based on a sliding time window, and determining that the tire is in a state of durability damage when the analysis results meet preset abnormal conditions, includes: For any moment within the sliding time window, calculate the Euclidean norm of the residual vector to obtain the damage score at the current moment; When the damage score at the current moment exceeds a preset safety threshold, an anomaly count marker is generated; Count the total number of anomaly counters within the sliding time window; When the total number exceeds a preset counting threshold, the tire is determined to be in a state of durability damage.

6. The method for identifying internal tire durability damage according to claim 1, characterized in that, The acquisition of dynamic response signals of the tire during operation, collected by a sensing unit arranged on the inner wall of the tire, includes: During tire rolling, the acceleration signal of the inner wall of the tire is received synchronously by all the sensing units; wherein the sensing units are arranged at intervals along the circumferential direction of the inner wall of the tire.

7. A tire internal durability damage identification system, characterized in that, include: The acquisition module is used to acquire the dynamic response signals of the tire during operation collected by the sensing unit arranged on the inner wall of the tire. The feature extraction module is used to extract features from the dynamic response signal and construct a current feature vector containing multidimensional features; The prediction module is used to input the current feature vector into the benchmark prediction model and obtain the predicted feature vector output by the benchmark prediction model; wherein, the benchmark prediction model includes a mechanism module for characterizing the physical properties of the tire and a data-driven module for compensating for nonlinear errors, and the benchmark prediction model is obtained by training the data-driven module with training samples, wherein the training samples are historical feature vectors under normal tire operating conditions; The residual calculation module is used to calculate the difference between the current feature vector and the predicted feature vector, and generate a residual vector. The discrimination module is used to perform continuous analysis on the statistical characteristics of the residual vector based on a sliding time window, and determine that the tire is in a state of durability damage when the analysis results meet preset abnormal conditions.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it implements the tire internal durability damage identification method as described in any one of claims 1 to 6.

9. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the tire internal durability damage identification method as described in any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by the processor, it implements the tire internal durability damage identification method as described in any one of claims 1 to 6.

Images