Tire wear depth intelligent analysis method and system

Abstract

The invention discloses a tire wear depth intelligent analysis method and system, and relates to the technical field of tire detection. According to the method, the resolution of the image sensor is automatically adjusted to adapt to different environment temperatures, and the K-means clustering algorithm and the support vector machine regression model are combined, so that the real-time performance and accuracy of tire tread image data acquisition are improved, the data processing efficiency is optimized, and the real-time performance of tire tread image data acquisition is improved. A complex model capable of accurately mapping the wear rate and the change of the feature points is established, the model has good generalization ability and can adapt to various tire types and use environments, the contradiction between high precision and real-time performance, data processing efficiency and accuracy and model complexity and generalization ability is effectively solved, and the real-time performance of the model is improved. Accurate analysis and prediction of the tire wear depth are achieved, and the intelligent level of tire monitoring is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tire detection, and specifically to an intelligent analysis method and system for tire wear depth. Background Art

[0002] Tire wear detection systems mainly provide tire wear monitoring for consumers, improve driving safety, prevent accidents caused by excessive tire wear, and provide customers with a more comprehensive tire maintenance solution. However, during the implementation of tire wear detection systems, a series of technical contradictions are faced, which directly affect the system's intelligent analysis ability for tire wear depth and the prediction accuracy of service life and maintenance requirements. First, there is a contradiction between high precision and real-time performance in the acquisition of tire tread depth information. High-precision acquisition requires the use of high-resolution sensors, but this will significantly increase data processing time and affect real-time performance. Low-resolution sensors can quickly obtain data, but the accuracy is insufficient and it is difficult to accurately reflect wear details. Secondly, there is a contradiction between the amount of data and processing efficiency during data processing. A large amount of data generated by high-precision sensors needs to be processed by efficient algorithms, otherwise it will cause system response delays. However, efficient algorithms often simplify the data processing process and will ignore key wear characteristics, affecting analysis accuracy. Moreover, there is a contradiction between model complexity and generalization ability between wear depth analysis and tire service life prediction. Complex models can analyze wear conditions more accurately, but have poor generalization ability and are difficult to adapt to different tire types and usage environments. Simple models have strong generalization ability, but the analysis accuracy is insufficient and it is difficult to accurately predict service life. Summary of the Invention

[0003] The purpose of the present invention is to provide an intelligent analysis method and system for tire wear depth, and to provide effective technical means for the technical contradictions in tire wear detection systems, so as to improve the intelligent analysis ability and prediction accuracy of the system.

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] The present application provides an intelligent analysis method for tire wear depth, including the following steps:

[0006] Collect tread image data at different ambient temperatures through an image sensor with adjustable resolution, and then obtain a standardized initial tire tread image;

[0007] According to the standardized initial tire tread image, extract tread depth data points as multiple input samples, and perform clustering analysis through the K-means clustering algorithm to determine the number of clusters and the initial cluster centers, obtain several groups and center points, and obtain tread depth distribution information;

[0008] Obtain the vehicle driving speed again. When the vehicle driving speed is lower than 5 kilometers per hour, collect tire tread image data at the highest data collection frequency within 30 seconds. When the driving speed is higher than 120 kilometers per hour, collect tread image data at the lowest frequency within 300 seconds to obtain multiple tread images at different times within a period, and judge the tire wear condition during the period;

[0009] According to the tread depth distribution information and the tread images at different times, extract the characteristic points of the tread depth change. Use the cubic spline interpolation fitting algorithm for the characteristic points to perform encrypted sampling processing on the characteristic points to obtain the characteristic point depth change data;

[0010] According to the tread depth distribution information and the characteristic point depth change data, establish a wear model using the support vector machine regression algorithm. Input the tread depth distribution information and the characteristic point depth change data into the training wear model to determine the mapping relationship between the wear rate and the characteristic point change;

[0011] Obtain the three-dimensional data of the road surface roughness, environmental temperature data, and driving behavior data, construct the tire three-dimensional input, and then transmit the tire three-dimensional input to the wear model to obtain the corrected wear rate;

[0012] According to the corrected wear rate and the initial tire depth, judge the current remaining service life of the tire according to the pre-established standard including the wear limit. When the current remaining service life of the tire is less than 300 hours, immediately start the tire replacement program to obtain a replacement reminder message.

[0013] Furthermore, obtain the standardized initial image of the tire tread, specifically including:

[0014] Obtain the real-time data stream of the environmental temperature and transmit it to the control unit. Judge the environmental temperature according to the real-time data stream of the environmental temperature. When the data stream represents that the environmental temperature value is lower than zero degrees Celsius, the control unit transmits the first control signal to the image sensor;

[0015] After receiving the first control signal, the image sensor adjusts to the highest resolution mode to collect the first tire tread image data. Judge the environmental temperature according to the real-time data stream of the environmental temperature. When the data stream represents that the environmental temperature value is higher than 35 degrees Celsius, the control unit transmits the second control signal to the image sensor;

[0016] After receiving the second control signal, the image sensor adjusts to the lowest resolution mode to collect the second tire tread image data. Judge the real-time data stream of the environmental temperature. When the data stream represents that the temperature is between zero degrees Celsius and 35 degrees Celsius, the control unit transmits the third control signal to the image sensor;

[0017] After receiving the third control signal, the image sensor calculates the resolution value using the resolution adjustment model to collect the third image data of the tire tread, and obtains the standardized initial image data.

[0018] Further, perform clustering analysis through the K-means clustering algorithm to determine the number of clusters and the initial cluster centers, obtain several groups and center points, and obtain the tread depth distribution information, specifically including:

[0019] According to the standardized tread image, preprocess the tread image, extract no less than 10,000 tread depth data points, and apply the K-means clustering algorithm to the set of no less than 10,000 tread depth data points, and input the set of tread depth data points;

[0020] According to the input set of tread depth data points, run the K-means clustering algorithm, output the clustering result, and judge that when the number of clusters or the initial cluster center does not meet the requirements of the clustering result, then adjust the number of clusters or the initial cluster center parameters to obtain new number of clusters or initial cluster center parameters;

[0021] Perform clustering operations again on the new number of clusters or initial cluster center parameters through the adjusted number of clusters or initial cluster center parameters to obtain an updated clustering result, and perform grouping information. According to the updated clustering result, output the grouping information and each center point to determine the tread depth distribution information.

[0022] Further, obtain multiple tread images at different time periods within a time period, and judge the tire wear condition during the time period, specifically including:

[0023] Obtain the real-time driving speed of the vehicle and continuously monitor it, with an interval of 1 second. When the detected driving speed is lower than 5 kilometers per hour, the acquisition control instruction is triggered, and the camera obtains the original tread image sequence at the current moment at a frequency of 30 frames per second within 30 seconds, and obtains a sequence containing 900 frames of images;

[0024] When the detected driving speed is higher than 120 kilometers per hour, another acquisition control instruction is triggered, and the camera obtains the initial tread image sequence at the current moment at a frequency of 1 frame per second within 300 seconds, and obtains a sequence containing 300 frames of images;

[0025] For each frame of image in the original tread image sequence and the initial tread image sequence, calculate the variance of the gray values of the eight-neighbor pixels inside each frame of tread image using the eight-neighbor processing method respectively, and screen out the images with a variance value greater than 50 to obtain two different second tread image sequences respectively;

[0026] Using an edge extraction algorithm, extract the texture contour information in all the second tread image sequences to obtain all the edge information images, and determine the depth value and width value of the tread pattern on the current tire surface.

[0027] Using time difference technology to process all the edge information images, calculate the difference in the depth value of the tread pattern between the first edge information image at the start of the time period and the last edge information image at the end of the time period, to obtain the depth change value and width change value of the tread pattern.

[0028] When the depth change value of the tread pattern is less than 2 mm and the width change value of the tread pattern is less than 2 mm, it is determined that the wear condition of the tire during this time period is normal. When the depth change value of the tread pattern is greater than 5 mm and the width change value of the tread pattern is greater than 5 mm, it is determined that the wear condition of the tire during this time period is severe wear.

[0029] Furthermore, obtain the tread images collected at different time periods and the corresponding tread depth distribution information. For the tread depth distribution information of each time period, extract the feature points where the tread depth change exceeds the preset threshold of 1 mm; then according to the extracted feature points, use the cubic spline interpolation algorithm to fit the depth change trend of the feature points to obtain a fitting curve, determine a 5-mm distance centered on each feature point as the neighborhood range, and perform encrypted sampling on the tread depth data within this neighborhood range.

[0030] Judge whether the number of detailed feature points in the sampled neighborhood is not less than 1000. If so, save the sampled data; otherwise, adjust the neighborhood range and resample.

[0031] After completing the encrypted sampling within the neighborhood for each feature point, summarize the depth change data of all the detailed feature points; then analyze the wear conditions in different regions of the tread.

[0032] Furthermore, determine the mapping relationship between the wear rate and the change of feature points, specifically including:

[0033] Collect the three-dimensional image information of the tire tread, obtain the tread depth distribution map, extract no less than 1000 feature points from the tread depth distribution map, track these feature points, and obtain the set of feature points after the tire rotates.

[0034] Compare the original set of feature points with the set of feature points after rotation, calculate the depth change value of each feature point, obtain the set of feature point depth changes, calculate the average wear rate of the tire according to the set of feature point depth changes and the operating parameters, generate a training data set, use the set of feature point depth changes in the training data set as the input variable of the support vector machine regression algorithm, and the wear rate as the output variable to obtain the hyperparameters of the regression algorithm.

[0035] Use the support vector machine regression algorithm to iteratively process the above parameters to obtain the optimal hyperparameters, determine the tire wear model, and then input the new tread depth distribution and characteristic point change data into the determined tire wear model to predict the tire wear state. Adjust the model according to the gap between the model output data and the real data.

[0036] Further, before constructing the three-dimensional input of the tire and then transmitting the three-dimensional input of the tire to the wear model to obtain the corrected wear rate, it includes:

[0037] Obtain the road surface information, ambient temperature, and driving behavior data including vehicle speed data, braking data, and turning data at a specific time in the time series, and determine whether the ambient temperature exceeds the glass transition temperature of the tire rubber. When the ambient temperature is lower than the glass transition temperature, the data is not updated.

[0038] Fuse all the information in the time series into the initial three-dimensional input information through multi-sensor data fusion technology. According to the tire model included in the initial three-dimensional input information, determine the initial rubber hardness value of the specific tire, and reconstruct the ambient temperature data in the time series for the data with the ambient temperature higher than the glass transition temperature. After fusing with the initial three-dimensional input information, it is used as the information matrix.

[0039] Obtain the rubber hardness in this time period through the first preset model according to the change rate of each group of data in the information matrix; calculate the road surface roughness value according to the obtained rubber hardness value and the road surface information in the time series through the second preset model, and obtain all the weight coefficients of the third preset model.

[0040] Further, construct the three-dimensional input of the tire, and then transmit the three-dimensional input of the tire to the wear model to obtain the corrected wear rate, specifically including: after obtaining multiple values of the road surface roughness, calculate the range and mean fusion value of the multiple values, and determine the first intermediate value of the contact area and pressure between the tire and the ground.

[0041] According to the first intermediate value, train the model on the rubber hardness value of the fusion value and the ambient temperature in the time series through the support vector machine algorithm to obtain the second intermediate value constructed based on the driver's driving behavior characteristics; then obtain the weighted average of the change rate of the driving behavior data according to the time series, and determine whether the average value is higher than the set threshold. When it is higher than the threshold, all the second intermediate values are updated, otherwise they are not updated.

[0042] Construct a three-dimensional input matrix for tire wear based on the latest obtained second intermediate value and the obtained information matrix; then calculate the input values at different time points under different environments, road conditions, and driving behaviors, and use a long short-term memory network model for multiple trainings to obtain the corresponding tire wear rates under different input value conditions, and then make predictions. When the obtained input value is negative, directly calculate the prediction result from the previous non-negative input value;

[0043] Judge whether each input value exceeds the preset range compared with the prediction result. When it exceeds the preset range, readjust all weight coefficients of the third preset model;

[0044] Update the predicted results of the tire wear rate that do not exceed the preset range to the original wear model, and fuse the updated tire wear rate with other environmental factors through a deep reinforcement learning algorithm and perform non-linear regression to obtain the final corrected wear rate result, and obtain the corrected tire wear rate data under different environments for environmental temperature, road surface roughness, and driving behavior data.

[0045] Furthermore, judge the current remaining service life of the tire according to the pre-established standard including the wear limit, specifically including:

[0046] Obtain the initial depth data of the tire and the preset wear limit standard threshold, and use them as the basic parameters for calculating the remaining service life of the tire. Then, collect the wear data of the tire in real time through a sensor, and calculate and correct the wear rate according to a certain time window to obtain the current wear rate parameter;

[0047] Use the corrected wear rate, the initial depth of the tire, and the wear limit standard threshold as inputs, and use a linear regression algorithm to establish a mathematical model of tire wear and service time to predict the remaining service life of the tire;

[0048] Judge whether the calculated remaining service life of the tire is lower than the preset critical value. When it is lower than the critical value, trigger the tire replacement reminder program. In the tire replacement reminder program, match and recommend replacement tire products according to the tire model and specification information, and generate a replacement reminder message.

[0049] The present invention provides a tire wear depth intelligent analysis system for implementing a tire wear depth intelligent analysis method, including:

[0050] An image acquisition module, which acquires tread image data at different environmental temperatures through an image sensor with adjustable resolution, and then obtains a standardized initial image of the tire tread;

[0051] The depth data extraction and analysis module extracts tread depth data points from the standardized tread image, uses the K-means clustering algorithm to perform clustering analysis on the depth data points, determines the number of clusters and the initial cluster centers, adjusts the clustering parameters until the requirements are met, outputs the grouping information and the center points, and determines the tread depth distribution information;

[0052] The speed monitoring and judgment module monitors the vehicle driving speed, adjusts the image acquisition frequency according to the speed, and judges the tire wear condition through image analysis;

[0053] The feature point extraction and interpolation fitting module extracts the feature points of the tread depth change, uses the cubic spline interpolation algorithm to fit the depth change trend of the feature points, and performs dense sampling around the feature points to obtain the depth change data of the detailed feature points;

[0054] The wear prediction model construction module uses the support vector machine regression algorithm to establish a wear model, inputs the tread depth distribution information and the depth change data of the detailed feature points, and determines the mapping relationship between the wear rate and the feature point change;

[0055] The data fusion and wear correction module obtains the road surface roughness, environmental temperature and driving behavior data, constructs a three-dimensional input of the tire, transmits the three-dimensional input to the wear model, and obtains the corrected wear rate;

[0056] The tire life prediction module judges the current remaining service life of the tire according to the corrected wear rate and the initial depth of the tire. When the predicted service life is less than 300 hours, it starts the tire replacement program and reminds the user to replace the tire.

[0057] The beneficial effects of the present invention are as follows:

[0058] An image sensor with adjustable resolution is used to collect tire tread image data. By automatically adjusting the resolution of the sensor according to the environmental temperature to adapt to different data collection requirements, when the environmental temperature is below zero degrees Celsius, the sensor collects images at the highest resolution to capture more details; while when the environmental temperature is above 35 degrees Celsius, the sensor collects images at the lowest resolution to quickly obtain data, solving the contradiction between high-precision collection and real-time performance. Such an intelligent adjustment mechanism ensures that optimized initial images can be obtained under different conditions while maintaining real-time performance;

[0059] Efficient algorithms are used to process the large amount of data generated by high-precision sensors. By using the K-means clustering algorithm to perform clustering analysis on the extracted tread depth data points, the number of clusters and the initial cluster centers are determined, so as to obtain the tread depth distribution information, solving the contradiction between the data volume and the processing efficiency in the data processing process, not only improving the data processing efficiency, but also optimizing the clustering accuracy by adjusting the clustering parameters to ensure the accuracy of the analysis;

[0060] The wear model is established by using the support vector machine regression algorithm. By inputting the tread depth distribution information and the depth change data of the detail feature points into the training wear model, the mapping relationship between the wear rate and the change of the feature points is determined. The optimal hyperparameters are found through iterative processing, and the tire wear model is determined. Then, the model is adjusted according to the gap between the model output and the real data to improve the prediction accuracy, so as to solve the contradiction between the model complexity and the generalization ability in the wear depth analysis and the tire service life prediction. This method not only considers the complexity of the model but also ensures the generalization ability of the model, enabling it to adapt to different tire types and usage environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] For better understanding and implementation, the technical solutions of this application will be described in detail below with reference to the drawings.

[0062] Figure 1 It is a schematic flowchart of a method for intelligent analysis of tire wear depth provided in Embodiment 1 of this application;

[0063] Figure 2 It is a schematic flowchart of a method for intelligent analysis of tire wear depth provided in Embodiment 1 of this application to obtain a standardized initial image of the tire tread;

[0064] Figure 3 It is a schematic flowchart of a method for intelligent analysis of tire wear depth provided in Embodiment 1 of this application to obtain the tread depth distribution information;

[0065] Figure 4 It is a schematic structural diagram of a system for intelligent analysis of tire wear depth provided in Embodiment 2 of this application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0066] To further elaborate on the technical means and effects adopted by the present invention to achieve the predetermined invention purpose, the exemplary embodiments will be described in detail herein, and the examples are shown in the drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. On the contrary, they are merely examples of methods and systems consistent with some aspects of this application as detailed in the appended claims.

[0067] The terms used in this application are only for the purpose of describing specific embodiments and are not intended to limit this application. The singular forms "a", "the", and "said" used in this application and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0068] The following will, in conjunction with the accompanying drawings and preferred embodiments, elaborate in detail on the specific implementation manners, features, and effects of the present invention.

[0069] Embodiment 1

[0070] Please refer to Figures 1 - 3 , this embodiment provides an intelligent analysis method for tire wear depth, including the following steps:

[0071] S1. Use an image sensor with adjustable resolution to collect tire tread image data. The pre-established default standard is that if the ambient temperature data is below zero degrees Celsius, the image sensor is controlled to collect image data at the highest resolution. If the ambient temperature is above 35 degrees Celsius, the image sensor is controlled to collect tread image data at the lowest resolution, obtaining a standardized initial image of the tire tread.

[0072] Further, obtaining the standardized initial image of the tire tread specifically includes:

[0073] S11. Obtain the real-time data stream of the ambient temperature and transmit it to the control unit. Judge the ambient temperature according to the real-time data stream of the ambient temperature. When the data stream represents that the ambient temperature value is below zero degrees Celsius, the control unit transmits a first control signal to the image sensor;

[0074] S12. After receiving the first control signal, the image sensor adjusts to the highest resolution mode to collect the first tire tread image data. Judge the ambient temperature according to the real-time data stream of the ambient temperature. When the data stream represents that the ambient temperature value is above 35 degrees Celsius, the control unit transmits a second control signal to the image sensor;

[0075] S13. After receiving the second control signal, the image sensor adjusts to the lowest resolution mode to collect the second tire tread image data. Judge the real-time data stream of the ambient temperature. When the data stream represents that the temperature is between zero degrees Celsius and 35 degrees Celsius, the control unit transmits a third control signal to the image sensor;

[0076] S14. After receiving the third control signal, the image sensor uses a resolution adjustment model to calculate the resolution value and collect the third tire tread image data, obtaining the standardized initial image data.

[0077] Specifically, the resolution of the image sensor is intelligently adjusted to collect tire tread image data, ensuring optimized initial images are obtained under different ambient temperatures. Specifically, when the ambient temperature is below zero degrees Celsius, the system automatically adjusts the sensor to the highest resolution to capture details, while when the temperature is above 35 degrees Celsius, the resolution is reduced to quickly obtain the image, so that standardized initial images of the tire tread can be obtained under different temperature conditions.

[0078] S2. Based on the standardized initial tire tread image, no less than 10,000 tread depth data points are extracted. The no less than 10,000 tread depth data points are used as multiple input samples, and clustering analysis is performed through the K-means clustering algorithm to determine the number of clusters and the initial cluster centers, obtaining several groups and center points, and obtaining the tread depth distribution information;

[0079] Further, clustering analysis is performed through the K-means clustering algorithm to determine the number of clusters and the initial cluster centers, obtaining several groups and center points, and obtaining the tread depth distribution information, specifically including:

[0080] S21. According to the standardized tread image, preprocess the tread image, extract no less than 10,000 tread depth data points, and apply the K-means clustering algorithm to the set of no less than 10,000 tread depth data points, inputting the set of tread depth data points;

[0081] S22. According to the input set of tread depth data points, run the K-means clustering algorithm, output the clustering result, and judge that when the number of clusters or the initial cluster centers do not meet the requirements of the clustering result, adjust the parameters of the number of clusters or the initial cluster centers to obtain new parameters of the number of clusters or the initial cluster centers;

[0082] S23. Through the adjusted parameters of the number of clusters or the initial cluster centers, perform clustering operations on the new parameters of the number of clusters or the initial cluster centers again to obtain an updated clustering result, and perform grouping information. According to the updated clustering result, output the grouping information and each center point to determine the tread depth distribution information.

[0083] Specifically, through in-depth analysis of the standardized tire tread image, at least 10,000 tread depth data points are extracted, and the K-means clustering algorithm is used to classify these data points to determine the distribution of tread wear. This process includes preprocessing the tread image, executing the clustering algorithm, evaluating the clustering result, and adjusting the clustering parameters when necessary to optimize the clustering accuracy. Finally, the grouping information and the center points are output to accurately obtain the depth distribution information of tire wear, providing data support for further wear analysis and prediction.

[0084] S3. Then obtain the vehicle driving speed. When the vehicle driving speed is lower than 5 kilometers per hour, collect tire tread image data at the highest data acquisition frequency within 30 seconds. When the driving speed is higher than 120 kilometers per hour, collect tread image data at the lowest frequency within 300 seconds to obtain multiple tread images at different times within a period of time, and judge the tire wear condition during the period;

[0085] Further, obtain multiple tread images at different times within a period of time, and judge the tire wear condition during the period, specifically including:

[0086] Obtain the real-time driving speed of the vehicle and continuously monitor it, with an interval of 1 second. When the detected driving speed is lower than 5 kilometers per hour, the acquisition control instruction is triggered, and the camera obtains the original tread image sequence at the current moment at a frequency of 30 frames per second within 30 seconds, obtaining a sequence containing 900 frames of images;

[0087] When the detected driving speed is higher than 120 kilometers per hour, another acquisition control instruction is triggered, and the camera obtains the initial tread image sequence at the current moment at a frequency of 1 frame per second within 300 seconds, obtaining a sequence containing 300 frames of images;

[0088] For each frame of image in the original tread image sequence and the initial tread image sequence, respectively use the eight-neighborhood processing method to calculate the variance of the gray values of the eight-neighborhood pixels inside each frame of tread image, and screen out the images with a variance value greater than 50, respectively obtaining two different second tread image sequences;

[0089] Use the edge extraction algorithm to extract the texture contour information in all second tread image sequences, obtain all edge information images, and determine the depth value and width value of the tread pattern on the current tire surface;

[0090] Use the time difference technology to process all edge information images, calculate the difference in the depth value of the tread pattern between the first frame of edge information image at the start of the time period and the last frame of edge information image at the end of the time period, obtaining the depth change value and width change value of the tread pattern;

[0091] When the depth change value of the tread pattern is less than 2 millimeters and the width change value of the tread pattern is less than 2 millimeters, it is determined that the wear condition of the tire during this time period is normal. When the depth change value of the tread pattern is greater than 5 millimeters and the width change value of the tread pattern is greater than 5 millimeters, it is determined that the wear condition of the tire during this time period is severe wear.

[0092] Specifically, by monitoring the driving speed of the vehicle, the acquisition frequency of the tire tread images is adjusted to ensure high-frequency image acquisition at low speed (below 5 km / h) and low-frequency image acquisition at high speed (above 120 km / h), thereby obtaining the tread image sequences under different driving states. Through the eight-neighborhood processing, edge extraction and time difference technology of these images, the depth and width changes of the tire tread can be accurately measured, and then the wear condition of the tire can be judged according to the change values. When the change values are small (both the depth and width changes are less than 2 millimeters), it is considered that the wear is normal, while when the change values are large (both are greater than 5 millimeters), it is determined to be severe wear, providing a scientific basis for the maintenance and replacement of the tire.

[0093] S4. According to the tread depth distribution information and the tread images at different time periods, feature points where the tread depth change exceeds a preset threshold of 1 mm are extracted. The cubic spline interpolation fitting algorithm is used for these feature points, and encrypted sampling is performed on the 5-mm distance range around these feature points to obtain depth change data of no less than 1000 detailed feature points;

[0094] Further, obtain the tread images collected at different time periods and the corresponding tread depth distribution information. For the tread depth distribution information of each time period, extract the feature points where the tread depth change exceeds the preset threshold of 1 mm; then, according to the extracted feature points, use the cubic spline interpolation algorithm to fit the depth change trend of the feature points to obtain a fitting curve, determine the 5-mm distance centered on each feature point as the neighborhood range, and perform encrypted sampling on the tread depth data within this neighborhood range;

[0095] Judge whether the number of detailed feature points in the sampled neighborhood is no less than 1000. If so, save the sampled data; otherwise, adjust the neighborhood range and resample;

[0096] After completing the encrypted sampling within the neighborhood for each feature point, summarize the depth change data of all detailed feature points; then analyze the wear conditions of different regions of the tread to provide data support for subsequent tread prediction and maintenance.

[0097] Specifically, by analyzing the tread depth distribution information and the tread images at different time periods, accurately extract the feature points where the depth change exceeds 1 mm, and use the cubic spline interpolation algorithm to fit the depth change trend of these feature points. Then, perform encrypted sampling within the 5-mm neighborhood range around each feature point to obtain depth change data of no less than 1000 detailed feature points. This process ensures in-depth analysis of the key areas of tire wear and provides accurate data support for tire wear prediction and maintenance.

[0098] S5. According to the tread depth distribution information and the depth change data of no less than 1000 detailed feature points, establish a wear model using the support vector machine regression algorithm. Input the tread depth distribution information and the depth change data of no less than 1000 detailed feature points into the training wear model to determine the mapping relationship between the wear rate and the change of feature points;

[0099] Further, determining the mapping relationship between the wear rate and the change of feature points specifically includes:

[0100] Collect the three-dimensional image information of the tire tread, obtain the tread depth distribution map, extract no less than 1000 feature points from the tread depth distribution map, track these feature points, and obtain the set of feature points after the tire rotates;

[0101] Compare the original set of feature points with the set of feature points after operation, calculate the depth change value of each feature point, obtain the set of feature point depth changes, calculate the average tire wear rate based on the set of feature point depth changes and operating parameters, generate a training data set, use the set of feature point depth changes in the training data set as the input variable of the support vector machine regression algorithm, and the wear rate as the output variable to obtain the hyperparameters of the regression algorithm;

[0102] Use the support vector machine regression algorithm to iteratively process the above parameters to obtain the optimal hyperparameters, determine the tire wear model, and then use the new tread depth distribution and feature point change data to input the determined tire wear model to predict the tire wear state, and adjust the model according to the gap between the model output data and the real data.

[0103] Specifically, by combining the tread depth distribution information and the depth change data of at least 1000 detailed feature points, an accurate tire wear model is constructed using the support vector machine regression algorithm (SVR). This model can learn and determine the mapping relationship between the wear rate and the feature point changes, so as to predict the tire wear state. By collecting the three-dimensional image information of the tire tread, extracting the feature points, and tracking the changes of these feature points after the tire runs, the depth change values of the feature points are calculated. These change values and the operating parameters are used together to calculate the average tire wear rate and generate a training data set. Subsequently, these data are input into the SVR model, and the optimal hyperparameters are found through iterative processing, and finally the tire wear model is determined. Using this model, the tire wear state can be predicted, and the model can be adjusted according to the gap between the model output and the real data to improve the prediction accuracy.

[0104] S6. Obtain the three-dimensional data of the road surface roughness, ambient temperature data, and driving behavior data, construct the three-dimensional tire input, and then transmit the three-dimensional tire input to the wear model to obtain the corrected wear rate;

[0105] Further, before constructing the three-dimensional tire input and then transmitting the three-dimensional tire input to the wear model to obtain the corrected wear rate, it includes:

[0106] Obtain the road surface information, ambient temperature, and driving behavior data including vehicle speed data, braking data, and turning data at a specific time in the time series, and judge whether the ambient temperature exceeds the glass transition temperature of the tire rubber. When the ambient temperature is lower than the glass transition temperature, the data is not updated;

[0107] All the information within the time series is fused into the initial three-dimensional input information through multi-sensor data fusion technology. According to the tire model contained in the initial three-dimensional input information, the initial rubber hardness value of a specific tire is determined, and the environmental temperature data higher than the glass transition temperature is reconstructed within the time series and fused with the initial three-dimensional input information to form an information matrix.

[0108] The rubber hardness within this time period is obtained through the first preset model based on the change rate of each group of data in the information matrix; through the second preset model, the road surface roughness value is calculated according to the obtained rubber hardness value and the road surface information within the time series, and all the weight coefficients of the third preset model are obtained.

[0109] Further, a three-dimensional tire input is constructed and then transmitted to the wear model to obtain the corrected wear rate, which specifically includes: after obtaining multiple values of the road surface roughness, the range and mean fusion value of the multiple values are calculated to determine the first intermediate value of the contact area and pressure between the tire and the ground.

[0110] Based on the first intermediate value, the rubber hardness value of the fusion value and the environmental temperature of the time series are model-trained through the support vector machine algorithm to obtain the second intermediate value constructed based on the driver's driving behavior characteristics; then the weighted average of the change rate of the driving behavior data is obtained according to the time series, and it is judged whether the average value is higher than the set threshold. When it is higher than the threshold, all the second intermediate values are updated, otherwise they are not updated.

[0111] According to the obtained latest second intermediate value and all the data of the information matrix obtained in the previous steps through the formula: L = f(T, R, B) = aT + bR + cB, a three-dimensional tire wear input matrix is constructed; L represents the tire wear rate, T represents the environmental temperature, R represents the road surface roughness, B represents the driving behavior data, and a, b, and c are their weight coefficients respectively.

[0112] The input values at different time points under different environments, road conditions, and driving behaviors are calculated, and the long short-term memory network model is used for multiple trainings to obtain the corresponding tire wear rates under different input value conditions, and then predictions are made. When the obtained input value is negative, the prediction result is directly calculated from the previous non-negative input value.

[0113] It is judged whether each input value exceeds the preset range compared with the prediction result. When it exceeds the preset range, all the weight coefficients of the third preset model are readjusted.

[0114] Based on the prediction result of the tire wear rate within the preset range, update it to the original wear model. Through the deep reinforcement learning algorithm, fuse the updated tire wear rate with other environmental factors and obtain the final corrected wear rate result after non-linear regression, so as to obtain the corrected tire wear rate data under different environments with environmental temperature, road surface roughness, and driving behavior data.

[0115] Specifically, by comprehensively analyzing the three-dimensional data of road surface roughness, environmental temperature, and driving behavior, a comprehensive tire wear input model is constructed, which not only considers the change in the hardness of tire rubber, but also calculates the rubber hardness and road surface roughness values under different conditions through a preset model, and then obtains the corrected wear rate. Specifically, it will determine whether to update the data according to whether the environmental temperature exceeds the glass transition temperature of the tire rubber, and then use the multi-sensor data fusion technology to fuse the road surface information, environmental temperature, and driving behavior data into three-dimensional input information. Through the training of the support vector machine algorithm and the long short-term memory network model, it can predict the tire wear rate under different input conditions, and adjust the model parameters according to the actual data to obtain a more accurate wear prediction result. Finally, it can provide the corrected tire wear rate data matching the environmental temperature, road surface roughness, and driving behavior data, providing a scientific basis for tire maintenance and replacement decisions.

[0116] S7. According to the corrected wear rate and the initial depth of the tire, judge the current remaining service life of the tire according to the pre-established standard including the wear limit. When the current remaining service life of the tire is less than 300 hours, immediately start the tire replacement program to obtain a replacement reminder message.

[0117] Furthermore, judging the current remaining service life of the tire according to the pre-established standard including the wear limit specifically includes:

[0118] Obtain the initial depth data of the tire and the preset wear limit standard threshold, and use them as the basic parameters for calculating the remaining service life of the tire. Then, collect the wear data of the tire in real time through a sensor, calculate and correct the wear rate according to a certain time window to obtain the current wear rate parameter;

[0119] Take the corrected wear rate, the initial depth of the tire, and the wear limit standard threshold as inputs, and use the linear regression algorithm to establish a mathematical model of tire wear and service time to predict the remaining service life of the tire;

[0120] Judge whether the calculated remaining service life of the tire is lower than the preset critical value (such as 300 hours). When it is lower than the critical value, trigger the tire replacement reminder program. In the tire replacement reminder program, match the recommended replacement tire products according to the tire model and specification information, and generate a replacement reminder message;

[0121] The replacement reminder information is pushed to the vehicle owner through various channels such as mobile applications and in-vehicle infotainment systems to remind the owner to replace the tires in a timely manner, ensuring driving safety. By establishing a database for tire use and replacement, information such as the use time, wear rate change, and replacement time of each tire is recorded. Through big data analysis, the tire life prediction model and replacement reminder strategy are optimized to improve the intelligence level of the system.

[0122] Specifically, by comprehensively considering the corrected tire wear rate, the initial depth of the tire, and the preset wear limit standard, the accurate prediction of the remaining service life of the tire is achieved. First, the initial depth data and wear limit threshold of the tire are collected, and then combined with the real-time collected wear data, a mathematical model of tire wear and use time is established using a linear regression algorithm to predict the remaining service life of the tire. When the calculated remaining service life is lower than the preset critical value (such as three hundred hours), the system will automatically trigger the tire replacement reminder program. This program can not only match and recommend replacement tire products according to the tire model and specification information, but also send replacement reminders to the vehicle owner through various channels (such as mobile applications and in-vehicle infotainment systems) to ensure driving safety. In addition, by establishing a database for tire use and replacement, information such as the use time, wear rate change, and replacement time of each tire can be recorded, and the tire life prediction model and replacement reminder strategy are continuously optimized through big data analysis to improve the intelligence level.

[0123] Embodiment 2

[0124] Please refer to Figure 4 , this embodiment provides a tire wear depth intelligent analysis system for implementing a tire wear depth intelligent analysis method, including:

[0125] An image acquisition module that acquires tire tread image data at different ambient temperatures through an image sensor with adjustable resolution, and then obtains a standardized initial image of the tire tread;

[0126] A depth data extraction and analysis module that extracts no less than 10,000 tire tread depth data points from the standardized tire tread image, uses the K-means clustering algorithm to perform clustering analysis on the depth data points, determines the number of clusters and the initial cluster centers, adjusts the clustering parameters until the requirements are met, outputs the grouping information and the center points, and determines the tire tread depth distribution information;

[0127] A speed monitoring and judgment module that monitors the vehicle driving speed, adjusts the image acquisition frequency according to the speed, and judges the tire wear condition through image analysis;

[0128] Feature point extraction and interpolation fitting module, which extracts feature points with the tread depth change exceeding 1 mm, uses the cubic spline interpolation algorithm to fit the depth change trend of the feature points, and performs dense sampling around the feature points to obtain no less than 1000 depth change data of detailed feature points;

[0129] Wear prediction model construction module, which uses the support vector machine regression algorithm to establish a wear model, inputs the tread depth distribution information and the depth change data of detailed feature points, and determines the mapping relationship between the wear rate and the change of feature points;

[0130] Data fusion and wear correction module, which obtains road surface roughness, ambient temperature and driving behavior data, constructs a three-dimensional input of the tire, transmits the three-dimensional input into the wear model, and obtains a corrected wear rate.

[0131] Tire life prediction module, which judges the current remaining service life of the tire according to the corrected wear rate and the initial depth of the tire. When the predicted service life is less than 300 hours, it starts the tire replacement program and reminds the user to replace the tire.

[0132] The above is only a preferred embodiment of the present invention, and does not impose any form of limitation on the present invention. Although the present invention has been disclosed as above with a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art can make some changes or modifications to the equivalent embodiments with equivalent changes within the scope of the technical solution of the present invention. However, as long as it does not depart from the content of the technical solution of the present invention, any brief modification, equivalent change and modification made to the above embodiments according to the technical essence of the present invention still fall within the scope of the technical solution of the present invention.

Claims

1. A tire wear depth intelligent analysis method, characterized by: The steps include: The image data of the tire tread under different ambient temperatures are collected by an image sensor with adjustable resolution, and then a standardized initial image of the tire tread is obtained; According to the standardized initial image of the tire tread, the tread depth data points are extracted and used as multiple input samples. The clustering analysis is performed through the K-means clustering algorithm to determine the number of clusters and the initial cluster center, obtain several groups and center points, and obtain the tread depth distribution information; Then obtain the vehicle's driving speed. When the vehicle's driving speed is less than 5 kilometers per hour, collect tire tread image data within 30 seconds. When the vehicle's driving speed is more than 120 kilometers per hour, collect tire tread image data within 300 seconds. Obtain multiple tread images at different time periods within a time period to determine the tire wear condition within the time period. According to the tread depth distribution information and the tread images at different time periods, the characteristic points of the tread depth change are extracted, and the cubic spline interpolation fitting algorithm is used to perform encrypted sampling processing on the characteristic points to obtain the characteristic point depth change data; Based on the tread depth distribution information and the feature point depth change data, a wear model is established using a support vector machine regression algorithm. The tread depth distribution information and the feature point depth change data are input into the training wear model to determine the mapping relationship between the wear rate and the feature point change. Obtain three-dimensional data of road roughness, ambient temperature data, and driving behavior data, construct a three-dimensional tire input, and then transmit the three-dimensional tire input to the wear model to obtain a corrected wear rate; Based on the corrected wear rate and the initial depth of the tire, the remaining service life of the tire is determined according to the pre-established standards including the wear limit. When the remaining service life of the tire is less than 300 hours, the tire replacement program is immediately started and a replacement reminder message is obtained.

2. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Obtain a standardized initial image of the tire tread, including: Acquire the real-time data stream of the ambient temperature and transmit it to the control unit, determine the ambient temperature according to the real-time data stream of the ambient temperature, and when the data stream indicates that the ambient temperature value is lower than zero degrees Celsius, the control unit transmits a first control signal to the image sensor; After receiving the first control signal, the image sensor is adjusted to the highest resolution mode to collect the first image data of the tire tread, and the ambient temperature is determined according to the real-time data stream of the ambient temperature. When the data stream indicates that the ambient temperature value is higher than thirty-five degrees Celsius, the control unit transmits a second control signal to the image sensor; After receiving the second control signal, the image sensor is adjusted to the lowest resolution mode to collect the second image data of the tire tread, and the real-time data stream of the ambient temperature is determined. When the data stream indicates that the temperature is between zero degrees Celsius and thirty-five degrees Celsius, the control unit transmits a third control signal to the image sensor; After receiving the third control signal, the image sensor uses a resolution adjustment model to calculate a resolution value to collect third image data of the tire tread, and obtains standardized initial image data.

3. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Cluster analysis is performed using the K-means clustering algorithm to determine the number of clusters and the initial cluster center, obtain several groups and center points, and obtain tread depth distribution information, including: Preprocessing the tread image according to the standardized tread image to extract no less than 10,000 tread depth data points, applying a K-means clustering algorithm to the set of no less than 10,000 tread depth data points to input the set of tread depth data points; According to the input tread depth data point set, the K-means clustering algorithm is run to output the clustering result. When the number of clusters or the initial cluster center does not meet the clustering result requirements, the number of clusters or the initial cluster center parameters are adjusted to obtain new number of clusters or initial cluster center parameters; By adjusting the number of clusters or the initial cluster center parameters, the clustering operation is re-performed on the new number of clusters or the initial cluster center parameters to obtain updated clustering results, and grouping information is performed. According to the updated clustering results, the grouping information and each center point are output to determine the tread depth distribution information.

4. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Obtain multiple tread images at different time periods within a time period to determine the tire wear condition within the time period, including: The real-time speed of the vehicle is obtained and continuously monitored at an interval of 1 second. When the speed is detected to be lower than 5 kilometers per hour, the acquisition control command is triggered, and the camera acquires the original tread image sequence at the current moment at a frequency of 30 frames per second within 30 seconds, obtaining a sequence of 900 frames of images; When the driving speed is detected to be higher than 120 kilometers per hour, another acquisition control instruction is triggered, and the camera acquires the initial tread image sequence at the current moment at a frequency of one frame per second within 300 seconds, obtaining a sequence containing 300 frames of images; According to each frame of the original tread image sequence and the initial tread image sequence, the eight-neighborhood pixel grayscale value variance of each frame of the tread image is calculated using the eight-neighborhood processing method, and the images with variance values ​​greater than fifty are screened out to obtain two different second tread image sequences; Using edge extraction algorithm, extract the texture profile information in all second tread image sequences, obtain all edge information images, and determine the depth value and width value of the tread pattern on the current tire surface; Using the time difference technology to process all edge information images, the difference in the depth value of the tread pattern between the first edge information image at the beginning of the time period and the last edge information image at the end of the time period is calculated to obtain the depth change value and width change value of the tread pattern; When the tread depth change is less than two millimeters and the tread width change is less than two millimeters, the tire is judged to be in normal wear condition during this time period. When the tread depth change is greater than five millimeters and the tread width change is greater than five millimeters, the tire is judged to be in severe wear condition during this time period.

5. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Acquire tread images collected at different time periods and corresponding tread depth distribution information, and extract feature points where the tread depth change exceeds a preset threshold of 1 mm for the tread depth distribution information of each time period; Then, based on the extracted feature points, the cubic spline interpolation algorithm is used to fit the depth change trend of the feature points to obtain a fitting curve, and the 5 mm distance centered on each feature point is determined as the neighborhood range, and the tread depth data is encrypted and sampled within the neighborhood range; Determine whether the number of detail feature points in the sampled neighborhood is not less than 1000. If yes, save the sampled data; otherwise, adjust the neighborhood range and resample; After completing the encrypted sampling in the neighborhood of each feature point, the depth change data of all detail feature points are summarized; then the wear conditions of different areas of the tread are analyzed.

6. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Determine the mapping relationship between wear rate and feature point changes, including: Collecting three-dimensional image information of the tire tread, obtaining a tread depth distribution map, extracting no less than 1,000 feature points in the tread depth distribution map, tracking these feature points, and obtaining a set of feature points after the tire is running; Compare the original feature point set with the feature point set after operation, calculate the depth change value of each feature point, obtain the feature point depth change set, calculate the average tire wear rate according to the feature point depth change set and the operation parameters, generate a training data set, use the feature point depth change set in the training data set as the input variable of the support vector machine regression algorithm, and use the wear rate as the output variable to obtain the regression algorithm hyperparameters; The above parameters are iteratively processed using the support vector machine regression algorithm to obtain the optimal hyperparameters and determine the tire wear model. The new tread depth distribution and feature point change data are then input into the determined tire wear model to predict the tire wear status. The model is adjusted based on the gap between the model output data and the actual data.

7. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Before constructing the tire 3D input and transferring the tire 3D input to the wear model to obtain the corrected wear rate, it includes: Obtain road surface information, ambient temperature, and driving behavior data including vehicle speed data, braking data, and turning data at a specific time in the time series, and determine whether the ambient temperature exceeds the glass transition temperature of the tire rubber. When the ambient temperature is lower than the glass transition temperature, the data will not be updated; All information in the time series is fused into initial three-dimensional input information through multi-sensor data fusion technology. The initial rubber hardness value of a specific tire is determined according to the tire model contained in the initial three-dimensional input information. For data with ambient temperature higher than the glass transition temperature, the ambient temperature data and the initial three-dimensional input information are reconstructed in the time series as an information matrix after being fused. The rubber hardness within the time period is obtained through the first preset model according to the change rate of each group of data in the information matrix; the road roughness value is calculated based on the rubber hardness value and the road surface information in the time series through the second preset model to obtain all the value coefficients of the third preset model.

8. The tire wear depth intelligent analysis method according to claim 7, characterized in that: Constructing a three-dimensional tire input, and then transmitting the three-dimensional tire input to the wear model to obtain a corrected wear rate, specifically comprising: after obtaining multiple values ​​of road roughness, calculating the range and mean fusion values ​​of the multiple values, and determining a first intermediate value of the contact area and pressure between the tire and the ground; According to the first intermediate value, the model training is performed on the rubber hardness value of the fused numerical value and the ambient temperature of the time series through the support vector machine algorithm to obtain the second intermediate value constructed based on the driver's driving behavior characteristics; then the weighted average of the driving behavior data change rate is obtained according to the time series, and it is determined whether the average value is higher than the set threshold. If it is higher than the threshold, all the second intermediate values ​​are updated, otherwise they are not updated; According to the latest second intermediate value and the information matrix obtained, a three-dimensional input matrix of tire wear is constructed; then the input values ​​at different time points under different environments, road conditions, and driving behaviors are calculated, and the long short-term memory network model is used for multiple training to obtain the corresponding tire wear rate under different input value conditions, and then prediction is performed, and the input values ​​at different time points are input. When the input value obtained is a negative value, the prediction result is directly calculated from the previous non-negative input value; Determine whether each input value exceeds a preset range compared with the prediction result, and if so, readjust all value coefficients of the third preset model; Based on the tire wear rate prediction results that do not exceed the preset range, the original wear model is updated. The updated tire wear rate is integrated with other environmental factors through a deep reinforcement learning algorithm and the final corrected wear rate result is obtained through nonlinear regression. The corrected tire wear rate data under different environments with ambient temperature, road roughness and driving behavior data are obtained.

9. The tire wear depth intelligent analysis method according to claim 1, characterized in that: Determine the current remaining service life of the tire based on pre-established standards including wear limits, including: Obtain the initial depth data of the tire and the preset wear limit standard threshold value, and use them as the basic parameters for calculating the remaining service life of the tire. Then collect the tire wear data in real time through the sensor, and calculate and correct the wear rate according to a certain time window to obtain the current wear rate parameter; The corrected wear rate, initial tire depth and wear limit standard threshold are used as inputs, and a mathematical model of tire wear and usage time is established using a linear regression algorithm to predict the remaining service life of the tire. It is determined whether the calculated remaining service life of the tire is lower than a preset critical value. If it is lower than the critical value, the tire replacement reminder program is triggered. In the tire replacement reminder program, the recommended replacement tire products are matched according to the tire model and specification information, and a replacement reminder message is generated.

10. A tire wear depth intelligent analysis system, used to implement a tire wear depth intelligent analysis method according to any one of claims 1 to 9, characterized in that: include: An image acquisition module collects tread image data at different ambient temperatures through an image sensor with adjustable resolution, and then obtains a standardized initial image of the tire tread; The depth data extraction and analysis module extracts tread depth data points from the standardized tread image, performs cluster analysis on the depth data points using the K-means clustering algorithm, determines the number of clusters and the initial cluster center, adjusts the clustering parameters until the requirements are met, outputs the grouping information and the center point, and determines the tread depth distribution information; Speed ​​monitoring and judgment module, which monitors the vehicle speed, adjusts the image acquisition frequency according to the speed, and judges the tire wear condition through image analysis; Feature point extraction and interpolation fitting module, which extracts the feature points of tread depth change, uses the cubic spline interpolation algorithm to fit the depth change trend of the feature points, performs encrypted sampling around the feature points, and obtains the depth change data of the detailed feature points; The wear prediction model building module uses the support vector machine regression algorithm to build a wear model, inputs the tread depth distribution information and the depth change data of the detail feature points, and determines the mapping relationship between the wear rate and the feature point change; The data fusion and wear correction module obtains road roughness, ambient temperature and driving behavior data, constructs a three-dimensional tire input, transmits the three-dimensional input to the wear model, and obtains the corrected wear rate; The tire life prediction module determines the current remaining service life of the tire based on the corrected wear rate and the initial depth of the tire. When the predicted service life is less than three hundred hours, the tire replacement program is started and the user is reminded to replace the tire.

Images