Vehicle Tire Dynamic Load Monitoring System and Method

Abstract

This invention relates to the field of tire load monitoring technology, specifically to a vehicle tire dynamic load monitoring system and method. The system includes a tire position distance monitoring device, a data acquisition and transmission device, and a host computer. The tire position distance monitoring device is used to collect tire deformation signals. The data acquisition and transmission device is used to convert the collected tire deformation signals into deformation data and transmit it to the host computer. The host computer acquires real-time tire pressure data and vehicle speed data via the vehicle's CAN bus or external sensors. The host computer is configured with an XGBoost prediction model, which uses tire deformation characteristic data, real-time tire pressure data, and vehicle speed data as raw input features to output predicted tire dynamic load values, achieving high-precision, single-value real-time monitoring of vehicle tire dynamic load.

Description

Technical Field

[0001] This invention relates to the field of tire load monitoring technology, and more specifically to a vehicle tire dynamic load monitoring system and method. Background Technology

[0002] With the increasing size and intelligence of vehicles, real-time monitoring of tire dynamic load is crucial for improving driving safety and extending tire life. However, existing tire dynamic load monitoring technologies have several shortcomings: First, traditional monitoring sensors are mostly metal foil strain gauges, which have limited range and poor mechanical adaptability, making them unable to cope with the large deformation working conditions of mining tires. They are prone to signal drift, mechanical failure, and other problems, and lack environmental robustness. Second, sensor deployment lacks scientific experimental screening and has not been optimized for the signal characteristics of different areas of the tire. Some monitoring areas are affected by factors such as airflow disturbances and high-speed tire rotation, resulting in low signal-to-noise ratios and difficulty in extracting effective features. Third, existing prediction models mostly use linear fitting or simple nonlinear algorithms, which are insufficient for analyzing the nonlinear physical mechanisms of tire deformation caused by the coupling of multiple factors such as load, tire pressure, and driving speed. Furthermore, they are mostly based on deformation data from a single area for regional calculations, which can easily lead to inconsistent prediction results and large monitoring errors.

[0003] Especially in complex and variable environments such as mining conditions, tires are affected by a combination of factors such as load, tire pressure, and speed during actual operation. Existing research is insufficient in modeling the multi-factor coupling mechanism of load-tire pressure-speed-deformation, making it difficult to meet the real-time monitoring needs of complex scenarios. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide a vehicle tire dynamic load monitoring system and method.

[0005] The present invention is achieved through the following technical solution: a vehicle tire dynamic load monitoring system, including a tire position distance monitoring device, a data acquisition and transmission device, and a host computer;

[0006] The tire position distance monitoring device is used to collect tire deformation signals;

[0007] The data acquisition and transmission device is used to convert the acquired tire deformation signals into deformation data and transmit them to the host computer;

[0008] The host computer obtains real-time tire pressure data and vehicle speed data through the vehicle's CAN bus or external sensors.

[0009] The host computer is equipped with an XGBoost prediction model. The XGBoost prediction model uses tire deformation feature data, real-time tire pressure data, and vehicle speed data as raw input features and outputs predicted values ​​of tire dynamic load.

[0010] Preferably, the tire position distance monitoring device is installed in the circumferential area of ​​the tire crown and the sidewall area;

[0011] The acquired tire deformation signals include:

[0012] The peak deformation value at the load application point is obtained from the tire position distance monitoring device in the circumferential region of the tire crown.

[0013] The stable tensile value after unloading is obtained from the tire position distance monitoring device in the sidewall area.

[0014] Preferably, the data acquisition and transmission device includes a multi-channel signal acquisition submodule, a signal processing submodule, a communication transmission submodule, and a power management submodule;

[0015] The multi-channel signal acquisition submodule is equipped with multiple communication interfaces for connecting to the tire position distance monitoring device and receiving the acquired tire deformation signals.

[0016] The signal processing submodule is used to perform voltage division measurement, amplification, and analog-to-digital conversion on the acquired tire deformation signal, converting it into effective deformation data;

[0017] The communication transmission submodule is used to upload valid deformation data to the host computer;

[0018] The power management submodule provides a stable power supply to the multi-channel signal acquisition submodule, signal processing submodule, and communication transmission submodule, ensuring the normal operation of each submodule.

[0019] Preferably, the XGBoost prediction model parameters are configured as follows: the number of decision trees is set to 2000, the learning rate is set to 0.01, and the maximum tree depth is set to 6.

[0020] Preferably, during training, the XGBoost prediction model uses a training database that includes tire static calibration data, dynamic calibration data, and the corresponding original input feature data.

[0021] Preferably, a tire equipped with a vehicle tire dynamic load monitoring system is mounted on the testing machine, and static calibration data is obtained through static experiments. The static experiments specifically include:

[0022] Set the air pressure gradient, and at each air pressure, control the testing machine to apply a vertical load. Gradually increase the load from 0 to the maximum set value, and record the deformation characteristic data in the static state.

[0023] Preferably, the dynamic calibration data acquisition process is as follows:

[0024] Based on the static experiment, a speed variable was added. While maintaining the tire pressure and load gradient, the tire was controlled to run under different speed gradients. After the data stabilized, the dynamic deformation characteristic data was recorded, which is the dynamic calibration data.

[0025] The vehicle tire dynamic load monitoring method, applied to the aforementioned vehicle tire dynamic load monitoring system, includes the following steps:

[0026] S1. Data acquisition: Obtain tire deformation data, real-time tire pressure data, and vehicle speed data.

[0027] S2. Data preprocessing: The tire deformation data is filtered, denoised, and feature extracted to generate deformation feature values.

[0028] S3. Perform tire dynamic load prediction. Take the deformation feature value, real-time tire pressure data, and vehicle speed data as the original input features and input them into the trained XGBoost prediction model. After the model is calculated, it outputs the predicted value of tire dynamic load.

[0029] Preferably, in step S2, the tire deformation data is divided into several independent detection intervals according to a preset time period. Each detection interval contains a preset number of continuous deformation data points. The top three deformation peaks in each detection interval are extracted. Valid feature data are screened by peak consistency judgment, and the average value of the valid feature data is calculated as the deformation feature value.

[0030] Preferably, the peak consistency judgment is as follows:

[0031] The difference between multiple deformation peaks extracted within the same detection interval is compared to determine whether the numerical difference between the multiple deformation peaks is within a preset threshold range.

[0032] If the maximum difference between multiple deformation peaks is less than or equal to a preset threshold, the peaks in the detection interval are determined to meet the consistency requirements and are considered a valid interval. The feature data in the valid interval are considered valid feature data.

[0033] If the maximum difference between multiple deformation peaks is greater than a preset threshold, the peaks in the detection interval are determined to not meet the consistency requirements, are invalid intervals, and are removed.

[0034] Compared with the prior art, the beneficial effects of the present invention are:

[0035] The vehicle tire dynamic load monitoring system and method proposed in this invention not only effectively adapts to large deformation conditions such as mining tires and improves robustness and stability in harsh environments such as mining areas, but also achieves effective feature extraction with high signal-to-noise ratio. It accurately analyzes the multi-factor nonlinear coupling effect of load, tire pressure, and speed on tire deformation, solves the problems of inconsistent calculation results and low prediction accuracy in traditional technologies, and finally realizes high-precision, single-value real-time monitoring of vehicle tire dynamic load.

[0036] This invention scientifically selects the circumferential and sidewall regions of the tire crown, which are sensitive to load changes and have stable signals, as effective monitoring points through combined static and dynamic experimental analysis. It eliminates the axial region of the tire crown, which is severely affected by airflow disturbances and has a low signal-to-noise ratio, thus reducing invalid data interference at the source. Simultaneously, it extracts deformation characteristic data for specific motion phases from the two effective monitoring areas (the peak deformation value at the load application point in the circumferential region of the tire crown and the stable tensile value in the sidewall region after the load is removed). Combined with real-time tire pressure and driving speed, a multi-dimensional feature system is constructed, ensuring the effectiveness and representativeness of the monitoring data and laying a data foundation for subsequent high-precision load prediction.

[0037] This invention abandons the traditional single linear fitting method and adopts the XGBoost ensemble learning algorithm to construct a unified tire dynamic load prediction model. It integrates four core features: peak circumferential deformation of the tire crown, stable tensile value of the tire sidewall, real-time tire pressure, and vehicle speed. Through weighted summation of decision tree groups and iterative fitting of residuals, it accurately establishes the nonlinear mapping relationship between multi-dimensional input features and tire dynamic load, effectively analyzing the complex physical mechanism of multi-factor coupling. This model breaks the limitations of regional calculation and directly outputs a unique load prediction value, solving the problem of inconsistent prediction results of traditional models. Experimental verification shows that the model's prediction determination coefficient R-squared value is greater than 0.98, and the average absolute percentage error is only 1.11%, achieving high-precision real-time monitoring of tire dynamic load under complex working conditions in mining areas.

[0038] This invention utilizes an indoor tire durability testing machine to design a multi-gradient static loading and dynamic rolling standardized calibration experiment, obtaining full-range training data covering different air pressure, load, and speed conditions.

[0039] This experimental method standardizes the data acquisition process, effectively decouples the influence weights of various environmental factors on tire deformation, provides a high-quality, high-confidence sample database for training machine learning models, significantly improves the generalization ability of the models, and ensures that the models can maintain stable prediction accuracy in complex scenarios. Attached Figure Description

[0040] Figure 1 This is a diagram showing the location of the tire position distance monitoring device inside the tire;

[0041] Figure 2 This is a schematic diagram of the deformation of the strain gauge in the circumferential region of the tire crown when the tire pressure is 1050 kPa and the speed is 25 km / h.

[0042] Figure 3 yes Figure 2 The corresponding heat map of tire deformation in a single cycle;

[0043] Figure 4 This is a schematic diagram of the deformation of the strain gauge in the tire sidewall area when the tire pressure is 1050 kPa and the speed is 25 km / h.

[0044] Figure 5 yes Figure 4 The corresponding heat map of tire deformation in a single cycle;

[0045] Figure 6 This is a schematic diagram of the deformation of strain gauges in the axial region of the tire crown when the tire pressure is 1050 kPa and the speed is 5 km / h.

[0046] Figure 7 This is a schematic diagram of the deformation of strain gauges in the axial region of the tire crown when the tire pressure is 1050 kPa and the speed is 25 km / h.

[0047] Figure 8 This is a block diagram of the data acquisition and transmission device.

[0048] Figure 9 It is a scatter plot of the predicted values ​​and the actual values ​​of different models;

[0049] Figure 10 This is a comparison chart of the predicted values ​​and actual values ​​from different models.

[0050] In the diagram: 1. Peripheral area of ​​the tire crown; 2. Sidewall area; 3. Axial area of ​​the tire crown. Detailed Implementation

[0051] 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 a part of the embodiments of the present invention, and not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0052] Example 1

[0053] This application proposes a vehicle tire dynamic load monitoring system, including a tire position distance monitoring device, a data acquisition and transmission device, and a host computer.

[0054] Tire positioning distance monitoring devices can be deployed in multiple dimensions inside the tire to collect tire deformation signals. (Refer to...) Figure 1This application uses the circumferential area 1 of the tire crown and the sidewall area 2 as effective monitoring points to ensure the signal-to-noise ratio and accuracy of dynamic monitoring data.

[0055] During vehicle operation, for the circumferential region 1 of the tire crown, peak deformation data was specifically collected when the tire positioning distance monitoring device rotated to the load application point (grounding center); for the sidewall region 2, stable tensile state data was specifically collected after the tire positioning distance monitoring device left the load application area as the tire rotated. These two deformation data at specific spatial locations and specific motion phases serve as valid inputs for subsequently constructing a dynamic load calculation model.

[0056] In this embodiment, the tire positioning distance monitoring device uses a flexible, high-precision tensile resistance strain gauge. It employs a modified polyurethane elastomer film as the flexible substrate, with an elongation at break greater than 1000%, enabling it to adapt to large deformation conditions such as those encountered by mining tires. A gallium-aluminum alloy filled with conductive ink is used to fabricate the sensitive grid, achieving a linear change in resistance signal with deformation. Compared to traditional metal foil strain gauges, the large range characteristic of the modified polyurethane substrate effectively solves the problems of limited measurement range and mechanical failure under large deformation conditions in tires.

[0057] During installation, in order to obtain effective deformation characteristics, the above-mentioned flexible high-precision tensile resistance strain gauges are installed in specific areas of the inner wall of the tire using a special adhesive, namely the circumferential area 1 of the tire crown and the sidewall area 2.

[0058] Reference Figures 2 to 5 This embodiment takes a tire pressure of 1050 kPa and a speed of 25 km / h as an example to analyze the strain data of the tire sidewall region and the circumferential region of the tire crown. It can be seen that the data in these two regions exhibit obvious periodic characteristics, and their period frequency corresponds to the tire speed, verifying the physical authenticity of the deformation signal. Furthermore, the time-domain characteristics of the signal can be divided into two significant intervals: the sparse data interval corresponds to the instant when the strain gauge passes through the load application point, at which time the deformation reaches the extreme value within the period; the dense data interval corresponds to the strain gauge being located in a non-direct contact area, where the deformation is relatively stable and the amplitude is small.

[0059] The circumferential region 1 of the tire crown is designed as follows: resistance strain gauges are attached circumferentially along the inner wall of the tire crown. This region is mainly used to capture tensile deformation at the tire contact patch, where the deformation shows a significant logarithmic correlation with the vertical load.

[0060] Reference Figure 2 and Figure 3Experimental data of the circumferential region of the tire crown and thermograms of tire deformation in a single cycle show that in one deformation cycle, the initial state is point A. The flattening of the curved surface under pressure and stretching at point D will cause the curved surfaces on both sides to be squeezed and contracted, causing point B to turn into a contracted state and further intensifying at point C. Point D directly bears the pressure and deformation reaches the maximum stretch. After point E is removed from the load, it turns into a contracted state and the contraction effect gradually weakens at point F. After being removed from the load, it returns to the initial state at point G and remains stable until the start of the next cycle.

[0061] Sidewall region 2 is designed as follows: resistance strain gauges are attached to the inside of the tire sidewall. This region is mainly used to capture radial compression and sidewall bulging deformation of the tire under load, and this location has a high linear response to load changes.

[0062] Reference Figure 4 and Figure 5 The experimental data of the sidewall region and the thermogram of tire deformation in a single cycle show that in one deformation cycle, the initial state is point a. Point d is compressed, causing the sidewall height of point d and the areas on both sides to decrease and enter a contraction state. The contraction amount is the same under different load conditions. The strain gauge contraction reaches the limit state. Other areas are in a stretching state due to the tire structure characteristics and internal pressure. From point b, which is close to the load area, to point c, the deformation state changes drastically from stretching to contraction. The contraction amount reaches the maximum at point d. Point e gradually moves away from the load area and the contraction amount decreases. From point f to point g, the initial stretching state is restored and remains stable until the start of the next cycle.

[0063] This application excludes the placement of strain gauges in the axial region 3 of the tire crown because this region is subject to significant airflow disturbance during dynamic driving, resulting in a low signal-to-noise ratio. Furthermore, in the axial region of the tire crown, the strain gauges attached axially are affected by airflow disturbances within the tire during tire rotation. While deformation data fluctuates relatively little at a speed of 5 km / h, at a speed of 25 km / h, the deformation data becomes irregular due to the airflow within the tire, and the higher the speed, the greater the impact on the data. Figure 6 and Figure 7 Therefore, data from this region was removed in the subsequent dynamic analysis.

[0064] Based on the above experimental analysis, this application identified two key deformation feature points as the basis for subsequent load monitoring. Point D in the circumferential region of the tire crown (i.e., the peak deformation at the load application point) and point g in the sidewall region (i.e., the stable tensile state after load removal) both exhibit excellent load sensitivity and signal stability under dynamic load conditions, meeting the technical requirements for high-precision monitoring.

[0065] The selection of point D is based on its extreme value characteristics and the clarity of its physical meaning within a single cycle. This point corresponds to the direct contact state between the tire and the ground, and there is a definite functional relationship between the deformation and the load. It is also less affected by speed, ensuring the comparability of data under different working conditions.

[0066] The selection of point g is based on its stability and representativeness throughout the entire rotation cycle. Located at the point of maximum tension in the tire sidewall region, although not the direct point of load application, it reflects the overall stress distribution and deformation pattern of the tire, providing important supplementary information for load monitoring. The deformation at point g is closely related to the tire's radial stiffness and tire pressure, effectively distinguishing different load conditions and providing reliable characteristic parameters for multi-parameter fusion load monitoring algorithms.

[0067] The data acquisition and transmission device is used to convert the tire deformation signals collected by the tire position distance monitoring device into valid deformation data and transmit it to the host computer. (Refer to...) Figure 8 The data acquisition and transmission device is designed as an integrated circuit board that can be installed inside the wheel hub. This enables high-precision signal acquisition and wireless transmission, significantly improving robustness and long-term operational stability in harsh environments such as mines.

[0068] The data acquisition and transmission device includes a multi-channel signal acquisition submodule, a high-precision signal processing submodule, a wireless transmission submodule, and a power management submodule. In this embodiment, the multi-channel signal acquisition submodule has multiple communication interfaces for connecting to the tire position distance monitoring device and receiving tire deformation signals (i.e., raw resistance signals) acquired by the device. In this embodiment, the communication interface is specifically a strain gauge interface. The high-precision signal processing submodule uses an N76E003AT20 microcontroller in conjunction with an LMV324IDR operational amplifier to perform voltage division measurement, amplification, and analog-to-digital conversion of the resistance signal, ultimately converting it into valid deformation data. The communication transmission submodule is responsible for transmitting the valid deformation data to the host computer in real time. In this embodiment, the communication transmission submodule uses an I8100 Bluetooth module. The power management submodule integrates an ME4054 chip to provide stable power to the multi-channel signal acquisition submodule, signal processing submodule, and communication transmission submodule, ensuring the normal operation of each submodule and guaranteeing long-term power supply for the system within the tire environment.

[0069] The host computer establishes a two-way data link with the data acquisition and transmission device via Bluetooth communication protocol. It can capture multi-channel resistance digital signals transmitted by the data acquisition and transmission device in real time, and convert the raw data into high-precision physical deformation based on the pre-calibrated strain-resistance characteristic curve through digital filtering and linearization processing algorithms. It provides a remote control interface to support the adjustment of key parameters such as sampling frequency, range, and channel selection. It can realize real-time graphical display of multi-threaded data and integrate a peak consistency detection algorithm to effectively extract feature point deformation.

[0070] The host computer also obtains real-time tire pressure data (kPa) and vehicle speed data (km / h) through the vehicle's CAN bus or external sensors.

[0071] To address the nonlinear physical mechanisms and complex multi-parameter interactions involved in tire deformation, and to avoid inconsistencies in load predictions caused by regional calculations, this application configures a prediction model based on the XGBoost machine learning algorithm (hereinafter referred to as the XGBoost prediction model) on the host computer to predict the dynamic load of vehicle tires.

[0072] XGBoost is an efficient ensemble learning algorithm based on gradient boosting decision trees. It belongs to the category of supervised learning. Its core is based on the idea of ​​boosting. It constructs multiple regression / classification decision trees through serial iteration. Each new tree focuses on fitting the residual between the prediction results of the preceding model and the true value. At the same time, it introduces regularization terms and uses second-order gradient information for optimization. It can effectively prevent overfitting while improving prediction accuracy. It is widely used in machine learning tasks such as regression and classification.

[0073] The XGBoost model performs a non-linear mapping by sequentially feeding the input feature vector into all decision trees. The predictions from each tree are then scaled using the learning rate as weights and summed in a weighted manner, ultimately outputting a single prediction. Its prediction function... The expression is:

[0074] ;

[0075] Where z represents the input feature vector; This represents the predicted tire dynamic load value; This indicates the total number of decision trees; Indicates the learning rate; This represents the predicted output of the t-th regression decision tree for the input feature vector; t represents the iteration round.

[0076] For XGBoost prediction model training, this embodiment utilizes an indoor tire durability testing machine (such as the TJR-2-OTR model) to conduct standardized experiments and construct a sample database for model training. In this embodiment, tires equipped with monitoring systems are mounted on the testing machine, and static calibration data is obtained through static experiments. The static experiments involve setting air pressure gradients (e.g., 850 kPa, 950 kPa, 1050 kPa, 1150 kPa, 1250 kPa), and at each air pressure, controlling the testing machine to apply a vertical load, gradually increasing the load from 0 to the maximum set value (e.g., 11200 kgf), recording the deformation characteristic data under static conditions to achieve static calibration.

[0077] Building upon static experiments, this application adds a speed variable to achieve dynamic calibration. Specifically, while maintaining air pressure and load gradients, the tire is controlled to run at different speed gradients (e.g., 0 to 30 km / h, in 5 km / h increments). After the data stabilizes, dynamic deformation characteristic data are recorded. Through these steps, a multi-condition, full-range training database is established, encompassing four dimensions: deformation, load, air pressure, and speed.

[0078] The relevant gradient settings and maximum settings can be adjusted according to the tire model.

[0079] The above data was divided into training and validation sets. The peak value of tire crown deformation, the stable tensile value of the tire sidewall, the real-time tire pressure, and the vehicle speed were used as input features, and the actual load applied by the testing machine was used as the output label to supervise the training of the XGBoost prediction model.

[0080] During training, a grid search method was used to optimize the key hyperparameters of the XGBoost prediction model. Considering the strong nonlinearity, multi-scale coupling, and complex interactions of the input features in tire dynamic load prediction, a large-scale tree ensemble was employed to enhance the model's ability to characterize local characteristics across different operating conditions. Numerous weak learners were progressively stacked using an additive model, effectively controlling variance while reducing model bias. Considering both computational resource constraints and model stability requirements, this embodiment determined the number of decision trees T to be 2000, the learning rate η to be 0.01, and the maximum tree depth to be 6. Regularization parameters γ and λ were introduced to constrain the complexity of the tree structure and prevent overfitting.

[0081] This application constructs a multi-dimensional feature input system. The peak deformation value of the tire crown circumferential region 1 at the load application point, the stable tensile value of the sidewall region 2 after load removal, the real-time tire pressure, and the vehicle speed are used as the original input features of the XGBoost prediction model. Based on the trained XGBoost prediction model, a unique predicted value of the tire dynamic load is output.

[0082] Example 2

[0083] Based on Example 1, this example proposes a method for monitoring dynamic load on vehicle tires, including the following steps:

[0084] S1. Data acquisition: Obtain tire deformation data, real-time tire pressure data, and vehicle speed data.

[0085] The tire deformation signal is collected by the tire position distance monitoring device, converted into valid deformation data by the data acquisition and transmission device and sent to the host computer. Real-time tire pressure data (kPa) and vehicle speed data (km / h) are obtained based on the vehicle CAN bus or external sensors (such as air pressure sensor and vehicle speed sensor).

[0086] S2. Data preprocessing: The tire deformation data is filtered, denoised, and feature extracted to generate deformation feature values.

[0087] For deformation data, the host computer removes high-frequency noise using a moving average filtering algorithm and converts the resistance value into physical deformation (mm) based on the calibrated resistance-deformation characteristic curve.

[0088] Furthermore, the host computer receives the original deformation data (deformation peak value, stable tensile value) of the tire crown circumferential and sidewall regions collected by the tire position distance monitoring device. After preprocessing such as segmentation, peak extraction, and multi-peak consistency judgment, the corresponding deformation feature values ​​of the tire crown circumferential region and the tire sidewall region are generated and used as the deformation-type original input features of the load prediction model.

[0089] Considering the multiple constraints of statistical validity, computational efficiency, and signal stability, this application adopts a data segmentation strategy, dividing tire deformation data into several independent detection intervals according to a preset time period. Each detection interval contains a preset number of continuous deformation data points. The top three deformation peaks in each detection interval are extracted, and effective feature data are screened by peak consistency judgment. The average value of the effective feature data is calculated as the deformation feature value.

[0090] The peak consistency judgment is as follows:

[0091] The difference between multiple deformation peaks extracted within the same detection interval is compared to determine whether the numerical difference between the multiple deformation peaks is within a preset threshold range.

[0092] If the maximum difference between multiple deformation peaks is less than or equal to a preset threshold, the peaks in the detection interval are determined to meet the consistency requirements and are considered a valid interval. The feature data in the valid interval are considered valid feature data.

[0093] If the maximum difference between multiple deformation peaks is greater than a preset threshold, the peaks in the detection interval are determined to not meet the consistency requirements, are invalid intervals, and are removed.

[0094] This embodiment uses a sampling frequency of 50Hz, acquiring data once every 20ms, which meets the technical requirements for high-fidelity signal acquisition. Simultaneously, the high sampling frequency provides sufficient data density for subsequent digital filtering, feature extraction, and time-domain analysis, ensuring the accuracy and flexibility of signal processing.

[0095] A 60-second experiment at a 50Hz sampling frequency will generate 3000 data points. To ensure that each detection interval contains sufficient samples for statistical analysis, and to avoid the influence of signal non-stationarity caused by excessively long intervals, this application divides the data sequence into fixed-length intervals of 300 data points, corresponding to a 6-second time window. From a statistical perspective, a 6-second time window at the highest speed condition (30km / h) contains approximately 10-15 complete rotation cycles. The three largest values ​​in each interval are collected, and their differences are compared. If the difference is less than 0.01mm, the average value is used as the feature point.

[0096] The deformation data sequence is uniformly denoted as ,in, It is the i-th real-time strain data (deformation peak or stable tensile value) collected from a single monitoring area (the circumferential area of ​​the tire crown or the sidewall area).

[0097] The top three values ​​within each detection interval are identified as follows:

[0098] ;

[0099] in, For the first A set of three peak points within a given interval The peak values ​​are, in order, the first, second, and third largest peak values ​​in the k-th interval.

[0100] Subsequently, a multi-peak consistency judgment is performed, and the judgment formula is as follows:

[0101] ;

[0102] If the above conditions are met, then the average value is taken as the characteristic point of the interval, denoted as . It is represented as follows:

[0103] ;

[0104] Finally, the average value of the characteristic points in all intervals is taken as the deformation characteristic value for this working condition, denoted as F, which is expressed as follows:

[0105] ;

[0106] in, This represents the number of valid intervals in the operating condition.

[0107] Multi-peak consistency judgment effectively eliminates the interference of random noise and outliers on feature extraction, ensuring the representativeness and stability of feature points. At the same time, the setting of the difference threshold (0.01mm) takes into account the sensor accuracy and system noise level, providing a reliable judgment standard for feature point selection.

[0108] After receiving the deformation data (tire crown deformation peak and sidewall stretch value) collected by the tire position distance monitoring device, the host computer preprocesses the data and generates the corresponding deformation feature values, which serve as the original input features for the XGBoost prediction model.

[0109] S3. Perform tire dynamic load prediction. Take the deformation feature value, real-time tire pressure data, and vehicle speed data as the original input features and input them into the trained XGBoost prediction model. After the model is calculated, it outputs the predicted value of tire dynamic load.

[0110] This application collects tire deformation data (peak tire crown deformation and sidewall stability tensile value), tire pressure and vehicle speed data in real time, processes the data, and inputs it into the XGBoost prediction model to output a unique predicted value of tire dynamic load that integrates multi-source feature information.

[0111] Example 3

[0112] This embodiment uses existing model regression prediction models (tire point distance monitoring devices are located in the circumferential region of the tire crown and the sidewall region respectively), neural network models, and the system and method of this application to monitor tire load, referring to... Figure 9 and Figure 10 The predicted values ​​in this application have a higher degree of consistency with the actual values ​​and less dispersion, resulting in better accuracy and stability in tire dynamic load monitoring.

[0113] Experimental results show that the prediction determination coefficient R-squared value of this application is greater than 0.98, and the mean absolute percentage error is only 1.11%, achieving high-precision monitoring of tire dynamic load.

[0114] The above description is merely an optional embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent structural transformations made using the content of the present invention under the concept of the present invention, or direct / indirect applications in other related technical fields, are included within the patent protection scope of the present invention.

Claims

1. A vehicle tire dynamic load monitoring system, characterized in that, Includes tire position distance monitoring device, data acquisition and transmission device and host computer; The tire position distance monitoring device is used to acquire tire deformation signals; the tire position distance monitoring device is deployed in the circumferential area (1) of the tire crown and the sidewall area (2). The tire position distance monitoring device includes a resistance strain gauge, and the circumferential area (1) of the tire crown is arranged by pasting the resistance strain gauge along the circumferential direction of the inner wall of the tire crown. The acquired tire deformation signals include: The peak deformation value of the load application point is obtained by the tire position distance monitoring device at the circumferential region (1) of the tire crown; The stable tensile value after unloading is obtained from the tire position distance monitoring device at the tire sidewall region (2); The data acquisition and transmission device is used to convert tire deformation signals into deformation data and transmit them to the host computer; The host computer obtains real-time tire pressure data and vehicle speed data through the vehicle's CAN bus or external sensors. The host computer is configured with an XGBoost prediction model. The XGBoost prediction model uses deformation feature values, real-time tire pressure data, and vehicle speed data as raw input features and outputs predicted values ​​of tire dynamic load. The deformation feature values ​​are generated after filtering, noise reduction, and feature extraction of tire deformation data.

2. The vehicle tire dynamic load monitoring system according to claim 1, characterized in that, The data acquisition and transmission device includes a multi-channel signal acquisition submodule, a signal processing submodule, a communication transmission submodule, and a power management submodule; The multi-channel signal acquisition submodule is equipped with multiple communication interfaces for connecting to the tire position distance monitoring device and receiving the acquired tire deformation signals. The signal processing submodule is used to perform voltage division measurement, amplification, and analog-to-digital conversion on the acquired tire deformation signal, converting it into deformation data; The communication transmission submodule is used to upload deformation data to the host computer; The power management submodule is used to provide power to the multi-channel signal acquisition submodule, signal processing submodule, and communication transmission submodule.

3. The vehicle tire dynamic load monitoring system according to claim 1, characterized in that, The XGBoost prediction model parameters are configured as follows: the number of decision trees is set to 2000, the learning rate is set to 0.01, and the maximum tree depth is set to 6.

4. The vehicle tire dynamic load monitoring system according to claim 1, characterized in that, During training, the XGBoost prediction model uses a training database that includes tire static calibration data, dynamic calibration data, and the corresponding raw input feature data.

5. The vehicle tire dynamic load monitoring system according to claim 4, characterized in that, Tires equipped with a vehicle tire dynamic load monitoring system are mounted on a testing machine, and static calibration data is obtained through static experiments. The static experiments are as follows: Set the air pressure gradient, and at each air pressure, control the testing machine to apply a vertical load. Gradually increase the load from 0 to the maximum set value, and record the deformation characteristic data in the static state.

6. The vehicle tire dynamic load monitoring system according to claim 5, characterized in that, The dynamic calibration data acquisition process is as follows: Based on the static experiment, a speed variable was added. While maintaining the tire pressure and load gradient, the tire was controlled to run under different speed gradients. After the data stabilized, the dynamic deformation characteristic data was recorded, which is the dynamic calibration data.

7. A method for monitoring dynamic load on vehicle tires, characterized in that, The application of the vehicle tire dynamic load monitoring system according to any one of claims 1-6 includes the following steps: S1. Data acquisition: Obtain tire deformation data, real-time tire pressure data, and vehicle speed data. S2. Data preprocessing: The tire deformation data is filtered, denoised, and feature extracted to generate deformation feature values. S3. Perform tire dynamic load prediction. Take the deformation feature value, real-time tire pressure data, and vehicle speed data as the original input features and input them into the trained XGBoost prediction model to output the final tire dynamic load prediction value.

8. The vehicle tire dynamic load monitoring method according to claim 7, characterized in that, In step S2, the tire deformation data is divided into several independent detection intervals according to a preset time. Each detection interval contains a preset number of continuous deformation data points. The top three deformation peaks in each detection interval are extracted. Valid feature data are selected by peak consistency judgment, and the average value of the valid feature data is calculated as the deformation feature value.

9. The vehicle tire dynamic load monitoring method according to claim 8, characterized in that, The peak consistency judgment is as follows: The difference between multiple deformation peaks extracted within the same detection interval is compared to determine whether the numerical difference between the multiple deformation peaks is within a preset threshold range. If the maximum difference between multiple deformation peaks is less than or equal to a preset threshold, the peaks in the detection interval are determined to meet the consistency requirements and are considered a valid interval. The feature data in the valid interval are considered valid feature data. If the maximum difference between multiple deformation peaks is greater than a preset threshold, the peaks in the detection interval are determined to not meet the consistency requirements, are invalid intervals, and are removed.

Images