The tread wear of automobile tires directly affects the safety of automobile driving. The national standard GB 7258-2017 "Technical Conditions for Safety of Motor Vehicle Operation" requires that the depth of the pattern on the tire crown of car tires should not be less than 1.6mm after wear, and other vehicle tires The depth of the pattern on the crown shall not be less than 3.2mm; the tire tread cannot expose the tire cord due to local wear.
According to statistics, 46% of traffic accidents on expressways are caused by tire failures. Obviously, the hidden danger of tire quality has become the "number one killer" in traffic accidents, and in traffic accidents related to tires, 25% of the tire-related traffic accidents are caused by tread wear. The depth of the tread is too shallow and lower than safe. Value. Therefore, among the many evaluation indicators of tires, the depth of the tread pattern and the form of wear are important factors that affect the driving safety of the car.
Tire wear process and wear mechanism
Cars rely on tires to drive on the road, and the tire pattern is in direct contact with the road. The friction generated by the tire tread block and the road is the source of power for driving, braking and steering of the car, so the degree of wear of the tire tread directly affects the safety of the car. In the use of tires, in addition to normal uniform wear, a series of factors such as tire quality, air pressure, four-wheel positioning, and driver operation will cause abnormal wear on the tire surface.
Tire wear is generally divided into the following 3 processes:
1.
Intermolecular interaction of surface materials
There are two ways of interaction between friction surfaces: mechanical and molecular. The mechanical action can be the direct contact between the two friction surfaces, that is, two-body wear; it can also be the inclusion of external abrasive particles between the two surfaces, that is, three-body wear. Molecular interactions include mutual attraction and adhesion of two surfaces.
2.
Elastoplastic deformation of surface material
During the friction process, affected by surface deformation, interface temperature and environmental conditions, the surface material will undergo mechanical, organizational, physical and chemical changes.
3.
Destruction of surface material
The main forms of damage are ploughing, tearing, fatigue damage, spalling and abrasion patterns.
According to rubber wear theory, tire wear can be divided into adhesion wear, fatigue wear, abrasive wear, degradation wear, curl wear and erosion wear. The wear process of tires is very complicated. The change of its surface material is the direct cause of tire wear. Wear is often the result of multiple mechanisms. Under normal wear conditions, the most common form of wear is abrasive wear. In addition, a series of factors such as road surface conditions (including whether the road surface is wet and slippery, road texture, temperature, etc.), tire load, tire structure, driving operation and other factors are also critical to tire wear.
Current status of tire tread wear detection technology
At present, the detection of tire wear mostly uses some technical means to obtain the depth information of each point of the tire tread, so as to evaluate and analyze the tire wear. There are many types of existing detection methods, which can be generally divided into two categories: contact detection and non-contact detection. The classification block diagram of tread wear detection methods is shown in Figure 1.
Figure 1 Classification block diagram of tread wear detection methods
Contact detection is also called mechanical detection. It uses a mechanical probe to scan the object to be tested to complete the detection. The advantage is that it is cheap and easy to operate. Human factors adversely affect detection accuracy and efficiency. The advantage of non-contact detection over contact detection is that it avoids contact with the object to be detected, making the detection method more flexible. Non-contact detection mainly includes visual detection and sensor detection.
1
Contact detection
At present, tire tread depth gauges are cheap and portable, and are favored by car owners and major auto repair shops. The appearance of direct-reading depth gauges and digital-display depth gauges is shown in Figure 2.
Figure 2 Appearance of direct reading depth gauge and digital display depth gauge
Researchers from Shanghai Ocean University combined the depth gauge to design a tire tread depth intelligent detection system based on the Internet of Things. It embeds the communication chip in the pattern depth gauge, and the data measured by the depth gauge pointer is processed by the embedded single-chip microcomputer, and finally displayed on the touch screen. The system can detect, collect, send, process and evaluate tire tread depth data, and provide tire maintenance suggestions through mobile terminals.
2
Non-contact detection
With continuous breakthroughs and innovations in image processing technology, optoelectronic technology, computer technology and other fields, non-contact detection technology has been developed. Since contact detection has the disadvantages of slow measurement speed, low efficiency, and low accuracy, non-contact detection has been recognized by the majority of enterprises and consumers due to a series of advantages such as fast measurement speed, wide application range, and automatic operation.
Tire texture feature detection
Figure 3 Flow chart of tire texture feature detection
Common methods for tire texture extraction include threshold segmentation, Tamura texture features, gray-level co-occurrence matrix, etc. The characteristics and application effects of several common methods are as follows:
01
Improved threshold segmentation
Method features: secondary threshold segmentation, the variance value is used as the initial segmentation, and the region labeling is used as the secondary segmentation.
Application effect: short operation time, clear contour segmentation, and realization of data dynamics.
02
Tamura texture features
Method characteristics: Contains 6 components to evaluate the image situation.
Application effect: Accurately obtain the relationship between tire wear and roughness and contrast.
03
Region extraction
Method features: directly locate the texture area of interest, reducing the amount of texture processing
Figure 4 Flow chart of texture-based region of interest algorithm
Application effect: rapid and accurate extraction of the region of interest, and good robustness to the extraction of the wear status of new and old tires.
04
Gray level co-occurrence matrix
Method features: Contains 14 feature values, and uses the frequency of gray values to appear at the same time for statistics to reflect the spatial information of the image gray.
Application effect: accurately obtain the relationship between tire wear and energy, entropy, and correlation.
05
Wavelet transform
Method characteristics: use two-level wavelet transform and high-frequency sub-band to reconstruct the image, and use SIFT feature selection to obtain texture.
Application effect: Reduce the extraction of SIFT feature points, prevent the loss of detailed information in the image, ensure accuracy, and improve efficiency.
06
Radon transform, DT-CWT
Method characteristics: It maintains a good analysis ability of local changes in the time domain, has displacement invariance, reduces data redundancy, and can perfectly reconstruct textures.
Figure 5 Radon transform and DT-CWT combined algorithm flow chart
Application effect: improve the accuracy of extraction and solve the problem of the impact of image rotation and translation on accuracy.
花纹深度检测
At present, most researches on tire tread depth detection use laser triangulation and binocular triangulation.
Laser triangulation is to use a beam of laser light emitted by a light source to irradiate the plane of the object to be measured, reflect it, and finally image on the detector. When the position of the surface of the object changes, the image formed by it also shifts correspondingly on the detector.
图6 激光三角法原理示意
The binocular triangulation method is to use the left and right cameras to observe the same target point from a certain distance, the difference in direction (that is, the change in parallax) and the proportional relationship between similar triangles to find the distance. When the displacement of the target point changes, the depth is obtained by the difference of the distance before and after the target point changes.
Figure 7 Schematic diagram of the principle of binocular triangulation
Common methods for tire pattern detection include monocular point laser, monocular line laser, binocular point laser, etc. The characteristics and application effects of several methods are as follows:
01
Monocular spot laser
Method features: use the cooperation of the mobile module and the encoder to complete the scanning of the tread by the spot laser.
Application effect: single point detection speed is fast, image processing is simple, but detection efficiency is low.
02
Binocular spot laser
Method features: double cameras collect light spots and add spot roundness evaluation to ensure effective light spots.
Application effect: single-point detection accuracy is high, binocular camera effectively avoids occlusion, but detection efficiency is low.
03
Monocular line laser
Method features: The laser line covers the tread section, which is easy to collect by camera.
Application effect: high detection efficiency and good robustness.
04
Binocular without laser
Method features: use the difference between the pattern groove and the tread brightness to calculate the height difference, that is, the pattern depth, through binocular ranging and SGM algorithm.
Sensor detection
With the development of sensor technology, the miniaturization and diversification of detection elements are increasingly applied to various fields. It can be used not only in the laboratory, but also for the detection of tire wear while the vehicle is running.
Tokyo Institute of Technology has proposed a flexible patch-type strain sensor that utilizes changes in capacitance. The sensor uses the capacitance change caused by the applied strain to accurately detect the deformation of the tire, and uses amplitude modulation for wireless measurement.
The University of Bologna in Italy has proposed a new method for continuous measurement of tire mechanical deformation. When the tire deformation causes a change in the steel wire spacing in the tire carcass, the impedance of the steel wire in this area also changes. The deformation of the tire is measured by measuring such changes in impedance.
Virginia Tech embeds optical fiber sensors in car tires to detect tire strain and at the same time monitor the occurrence of coasting.
Rutgers University, New Jersey, USA, uses a sensor based on polyvinylidene fluoride (PVDF) to measure the deformation of the tread. Two PVDF deformation sensors are connected to the inner surface of the rubber tire. Once the tire surface is deformed, the connected PVDF sensor will The piezoelectric effect is generated, and then the electrical signal is output to reflect the deformation of the tread.
The German Continental Group combines the tire tread depth warning function with the tire pressure monitoring system (TPMS), by continuously monitoring the vibration frequency generated by the contact between the tire and the road, and comparing the worn vibration frequency with the vibration frequency of the unworn tire to predict the tire The degree of wear. When the amount of wear reaches the set critical value, TPMS will stop providing the early warning function. Professor KANWAR also developed an intelligent tire system to monitor tire wear. Through experiments, he found out the relationship between the tire's vertical vibration frequency and tire wear. The more wear, the greater the corresponding vertical vibration frequency, so as to monitor tires. State of wear.
Duke University in the United States has cooperated with companies to develop a carbon nanotube (a tiny cylinder of carbon atoms with a diameter of only one billionth of a meter) sensor technology to measure the changes in tread depth on the millimeter scale. As shown in Figure 8, tire rubber and tread structure interfere with this so-called "fringe field", and this interference is measured by the electrical response of the electrodes, thereby determining the thickness of the tire above the sensor.
Figure 8 Schematic diagram of the working principle of the carbon nanotube sensor
Analysis of Difficulties in Research on Tire Wear Detection Technology
Tire wear detection technology is practically related to people's lives and has important research significance. The main difficulty of the research in this field is that the collected tire tread images are easily affected by various environmental factors (water stains, inclusions, etc.), and foreign objects on the tire surface must be cleaned before inspection.
For actual driving detection, too fast tire rotation speed will cause smear in the image collection, and the image collected by the camera is susceptible to interference from ambient light, which directly affects the detection accuracy.
If the sensor is implanted inside the tire, it will also bring a series of problems. For example, the implanted position during the tire building process will affect the accuracy of the sensor detection. In the tire manufacturing process, the carcass has to withstand high temperature and high pressure and other processes. The sensor will be damaged in this process, which is also a difficult point for research.
In addition, the degree of bonding between the chip and the tire will directly affect the quality of the tire. If the bonding is not good, the tire rubber will be detached, and the gap will gradually increase, eventually leading to the scrap of the tire.
The development trend of tire wear detection technology
With the continuous development of image processing technology, optical technology and computer technology, contact tire wear detection will inevitably be replaced by non-contact detection.
Smart tire
At present, mature smart tire technology is mainly used in tire pressure, temperature and friction state monitoring and recording, and it is rarely used in tire wear detection. By embedding sensors in the rubber at different positions in the tire tyre, the purpose of detecting the state of the tire is achieved. Due to the low stiffness of rubber and the high stiffness of traditional stress sensors, the fusion of the sensor and the tire is a major difficulty, and the structure of the tire needs to be specifically designed. The production process of smart tires is different from that of general mechanical parts, and the installation and calibration process of sensors is also troublesome, which leads to very high production costs of smart tires. In addition, the measurement of sensors is easily affected by the environment. Severe jitter will occur during high-speed rotation, which will affect the robustness of the sensor's measurement. Research and development of low-cost, long-life, strong anti-interference sensors and improved manufacturing processes are the primary tasks for the development of smart tires.
Intelligent detection
At present, most automobile tire tread detection is based on the tire at rest or at low speed. The detection efficiency is not high, and it is easily affected by the environment. Moreover, the detection device is relatively fixed and cannot be directly installed on the vehicle for real-time detection. Foreign Tire Profiles company introduced the handheld laser scanner Groove Glove. The device can accurately measure the depth of the tire tread in a few seconds. It is durable and portable. It can work anywhere within the WiFi signal coverage. At the same time, it can recognize the owner's license plate through the built-in camera, and obtain tires and location diagnosis. All the necessary data required, and finally provide customers with the strategy of tire replacement in the form of a report. However, the device also has disadvantages. Each scan can only obtain data of a small area, and cannot directly obtain the depth data of the entire tire tread. If the entire tire data is required, the device must be used for multiple scans, which cannot be realized in real time. Automatic detection.
With the development of technologies such as machine vision, image processing, and machine learning, intelligent inspection will inevitably become an indispensable method in future automobile inspection technology. More convenient and intelligent detection methods are bound to be the focus of future development. At the same time, with the rapid development of reverse engineering, the 3D point cloud reconstruction technology has attracted more and more attention in the field of inspection. The formation of 3D images of tires through 3D reconstruction technology and the visual classification and analysis of tire wear are future scholars. Explore the trend. Therefore, studying how to apply deep learning technology in the field of tire tread detection will be the task of current research and has important value in practical applications.
(1) It is necessary to improve the adaptability and robustness of non-contact tire detection under complex driving environment conditions, especially the impact of illumination on visual detection.
(2) Smart tire technology is more mature in tire pressure and temperature detection, but there is less research in the field of tire tread wear detection. At the same time, the detection accuracy of the chip is limited and it is not easy to install and debug.
(3) Accelerate the application of deep learning and machine learning in the field of tire wear detection, which can detect in the driving and provide real-time feedback during driving; combine point cloud technology and three-dimensional reconstruction technology to achieve visual detection, and can communicate with the Internet through wireless transmission technology. Combine data to classify, analyze, predict and alert tire wear.
