A smart monitoring and collaborative control system for tires of new energy vehicles

Abstract

This invention discloses an intelligent monitoring and collaborative control system for tires in new energy vehicles, comprising a tire data remote monitoring unit, a tire defect intelligent detection unit, a tire pressure-drive collaborative control unit, a power battery differential pressure diagnosis unit, and a cloud platform management unit. These units are interconnected via a CAN bus. The invention achieves real-time acquisition and remote uploading of tire status data through tire pressure and temperature sensors, and employs a lightweight multi-scale residual attention algorithm to accurately detect tire defects in X-ray images. Based on tire pressure adaptive model predictive control and a slip ratio fuzzy tuning algorithm, it optimizes drive torque distribution under low tire pressure conditions, stabilizing the slip ratio within a safe range while balancing energy consumption and driving stability. Simultaneously, it links with the power battery differential pressure for fault diagnosis and early warning. This system features high integration, fast control response, and vehicle-mountable deployment, effectively improving the driving safety, energy economy, and intelligent fleet management level of new energy vehicles.

Description

Technical Field

[0001] This invention relates to the field of intelligent safety control technology for new energy vehicles, specifically to an intelligent monitoring and collaborative control system for new energy vehicle tires that integrates remote tire status monitoring, intelligent tire defect detection, tire pressure-drive coordinated control, and power battery differential pressure linkage diagnosis. Background Technology

[0002] With the rapid development of the new energy vehicle industry, tires, as a core component for vehicle driving, load-bearing, and propulsion, directly affect vehicle safety, energy economy, and overall vehicle reliability. Currently, tire-related monitoring and control technologies still have significant limitations in new energy vehicle fleet management and vehicle control.

[0003] Most existing tire pressure monitoring systems can only display data locally inside the vehicle's cab, and the data cannot be transmitted externally. This makes it difficult for vehicle managers to remotely obtain tire pressure, tire temperature, and fault alarm information from outside the vehicle, and makes it impossible to uniformly manage the tire status of the fleet.

[0004] Tire manufacturing and maintenance inspections largely rely on manual visual inspection or general inspection methods, which have low accuracy in identifying weak textures and minor defects in tire X-ray images. Furthermore, existing algorithms generally require high computing power, making it difficult to achieve lightweight and real-time deployment on the vehicle side, resulting in insufficient efficiency and accuracy in defect detection.

[0005] The control strategy of the drive system of new energy vehicles usually does not take the real-time tire pressure status into the dynamic adjustment basis. Under low tire pressure conditions, the rolling resistance of the tires increases significantly, which leads to increased vehicle energy consumption, shortened driving range, and is also prone to safety problems such as braking deviation and decreased driving stability.

[0006] Currently, the tire monitoring system, drive control system, and power battery management system operate independently, without forming a vehicle-level collaborative safety management mechanism. This makes it impossible to achieve coordinated early warning and optimization of tire abnormalities, drive status, and battery pressure differences, which is insufficient to meet the intelligent and safe operation requirements of new energy vehicles.

[0007] To address the aforementioned technical issues, this invention proposes an integrated, lightweight intelligent monitoring and collaborative control system for new energy vehicle tires that possesses vehicle-wide collaborative capabilities. Summary of the Invention

[0008] The purpose of this invention is to overcome the shortcomings of the prior art and provide an intelligent monitoring and collaborative control system for tires of new energy vehicles.

[0009] To achieve the above objectives, the present invention adopts the following technical solution: A new energy vehicle tire intelligent monitoring and collaborative control system includes a tire data remote monitoring unit, a tire defect intelligent detection unit, a tire pressure-drive collaborative control unit, a power battery differential pressure diagnosis unit, and a cloud platform management unit. Each unit is interconnected through the vehicle's CAN bus, and the data is uploaded to the cloud platform management unit via the vehicle terminal, realizing remote monitoring of tire status, intelligent defect identification, adaptive adjustment of drive torque, and linkage diagnosis of battery differential pressure faults.

[0010] Furthermore, the tire data remote monitoring unit includes a tire pressure and temperature sensor, a CAN transceiver, and an on-board remote monitoring terminal; the CAN transceiver is integrated into the tire monitoring display and converts tire pressure, temperature, and alarm signals into CAN message broadcasts, which are then uploaded to the cloud platform by the on-board terminal at a fixed frequency.

[0011] Furthermore, the CAN message includes four types of signals: tire position, tire pressure, tire temperature, and alarm status. The message is sent once per second, and the vehicle terminal uploads the message to the cloud platform every 10 seconds.

[0012] Furthermore, the intelligent tire defect detection unit employs a multi-scale residual attention lightweight detection algorithm, which executes sequentially as follows: (1) Perform Gaussian noise filtering and scale normalization on the tire X-ray images to generate a standard 640×640 input image; (2) Using a lightweight residual convolutional network as the backbone, shallow edge features, mid-layer texture features, high-layer semantic features and global contour features are extracted sequentially through four layers of downsampling to form a four-level multi-scale feature map; (3) In the feature fusion stage, a dual-path polarization attention module is introduced to perform parallel weighted fusion of global mean pooling and max pooling on the channel dimension, and to perform adaptive localization enhancement of defect regions on the spatial dimension, and output the enhanced multi-scale fusion feature map. (4) The dynamic focusing boundary loss function is used to perform bounding box regression. The focusing coefficient is dynamically adjusted according to the defect area to complete the accurate detection, positioning and classification of three types of minor defects: open crown lines, sparse sidewall cords and impurities.

[0013] Furthermore, the lightweight residual convolutional network is composed of stacked multi-level depth-separable residual blocks. Each residual block contains three layers: 1×1 pointwise convolution, 3×3 depthwise convolution, and 1×1 pointwise convolution. During the training phase, an identity mapping residual branch is introduced for feature enhancement. During the inference phase, the multi-branch structure is fused into a single-path convolutional structure through structural reparameterization, reducing the computing power consumption of the vehicle terminal and realizing real-time inference.

[0014] Furthermore, the dual-path polarized attention module independently sets up a channel attention branch and a spatial attention branch. The channel branch is used to highlight the texture features of tire defects and suppress background noise, while the spatial branch is used to lock the geometric position of defects and enhance the boundary contour. The two outputs are multiplied element-wise to obtain the final attention-enhanced feature map.

[0015] Furthermore, the tire pressure-drive coordinated control unit incorporates a tire pressure adaptive model predictive control algorithm. This algorithm constructs a dual-objective optimization function for energy consumption and stability based on real-time tire pressure values, tire temperature, vehicle speed, and wheel speed signals. Using tire slip ratio, drive motor efficiency, and yaw rate deviation as constraints, it solves for the optimal drive torque distribution sequence within the next three control cycles through rolling time-domain optimization. Based on the degree of deviation in single tire pressure, it dynamically allocates the torque ratio of the left and right wheels / front and rear axles to achieve optimal synergy between energy consumption reduction and driving stability under low tire pressure conditions.

[0016] Furthermore, the tire pressure-drive coordinated control unit is equipped with a tire pressure-slip ratio fuzzy adaptive tuning algorithm; When a single tire pressure is detected to be 25% lower than the standard tire pressure, the low tire pressure mode is automatically triggered. The stability weight coefficient is increased and the energy consumption weight coefficient is decreased in real time through fuzzy rules, and the upper limit of the drive torque is linearly limited according to the tire pressure decay ratio. At the same time, the target value of the slip ratio is adaptively adjusted according to the real-time road surface adhesion coefficient, and the tire slip ratio is stably locked in the safe range of 10% to 15% to prevent slippage, deviation and excessive energy consumption.

[0017] Furthermore, the power battery differential voltage diagnostic unit collects the voltage of individual power battery cells in real time. When the voltage difference between individual cells exceeds the 100mV threshold, an alarm is triggered, and equalization and charging repair are performed on lithium iron phosphate batteries, while direct replacement is performed on ternary lithium batteries.

[0018] Furthermore, the cloud platform management unit adopts a distributed storage architecture and has functions such as data parsing, visualization, data anomaly verification, alarm information push, and historical data statistical analysis.

[0019] Beneficial effects of this invention: 1. Enables remote uploading and visual monitoring of tire pressure, temperature, and alarm signals, supporting unified fleet management and proactive early warning of anomalies.

[0020] 2. Employing a lightweight, original defect detection algorithm, it can be deployed on-vehicle devices, providing high accuracy and speed for detecting minute defects and weak texture defects in tire X-ray images.

[0021] 3. Through the tire pressure-drive coordinated control algorithm, torque distribution is automatically optimized under low tire pressure conditions, reducing energy consumption and improving driving stability.

[0022] 4. Enable linked diagnosis of tire condition and power battery differential pressure faults, build a vehicle-level safety system, and reduce operational risks and maintenance costs. Attached Figure Description

[0023] Figure 1: Schematic diagram of the working process of the present invention. Detailed Implementation

[0024] This system is applicable to new energy passenger vehicles, buses, trucks, and other vehicle types. The following is a detailed explanation of the overall architecture, the working process of each unit, and the overall system workflow. Example

[0025] This system includes a tire data remote monitoring unit, a tire defect intelligent detection unit, a tire pressure-drive coordinated control unit, a power battery differential pressure diagnostic unit, a cloud platform management unit, a CAN bus, and an on-board terminal.

[0026] The tire data remote monitoring unit collects tire pressure and temperature signals; the tire defect intelligent detection unit completes tire X-ray image defect identification; the tire pressure-drive coordination control unit adjusts drive torque according to tire pressure; the power battery differential pressure diagnostic unit monitors the consistency of battery cell voltage; all units are interconnected via CAN bus; the vehicle terminal uploads data to the cloud platform management unit to realize remote monitoring, fault alarm and collaborative management.

[0027] Preferably, the tire data remote monitoring unit includes a tire pressure and temperature sensor, a CAN transceiver, and an on-board remote monitoring terminal.

[0028] Tire pressure and temperature sensors are installed on the wheel rims to collect tire pressure and temperature data in real time. The sensors wirelessly transmit the signals to the in-vehicle display. The display integrates a CAN transceiver, which converts the signals into CAN bus messages and broadcasts them externally. The messages include tire position, pressure, temperature, and alarm status, and are sent once per second. The on-board remote monitoring terminal receives the messages and uploads them to the cloud platform management unit at a frequency of 10 messages per second. When abnormalities such as air leaks or high temperatures occur, real-time uploading is immediately triggered.

[0029] Preferably, the intelligent tire defect detection unit employs a multi-scale residual attention lightweight detection algorithm, the process of which is as follows: The first step is to perform Gaussian filtering and scale normalization on the tire X-ray image to generate a 640×640 standard image; The second step involves four downsampling operations using a lightweight residual convolutional network to extract four levels of feature maps: shallow edges, mid-level textures, high-level semantics, and global contours. The third step is to feed the data into the dual-path polarization attention module for feature enhancement. The fourth step is to use the dynamic focusing boundary loss function to complete the bounding box regression and output the defect category, location and confidence level.

[0030] The dual-polarization attention module is divided into a channel attention branch and a spatial attention branch.

[0031] The channel attention branch performs global mean pooling and max pooling on the feature map, and the weighted fusion highlights the texture of defects and suppresses the background; the spatial attention branch performs position modeling on the features, locks the defect region and strengthens the contour; the two outputs are multiplied element by element to obtain the final enhanced feature map, which improves the detection rate of weak textures and small defects.

[0032] Preferably, the tire pressure-drive coordinated control unit takes real-time tire pressure, vehicle speed, wheel speed, tire temperature, and road surface adhesion coefficient as inputs, and incorporates a tire pressure adaptive model predictive control algorithm and a tire pressure-slip ratio fuzzy adaptive tuning algorithm.

[0033] The system calculates tire pressure deviation in real time, constructs a dual-objective optimization function of energy consumption and stability, and uses slip ratio, motor efficiency, and yaw rate as constraints to solve the optimal torque distribution sequence for the next three control cycles in a rolling manner.

[0034] When a single tire pressure is detected to be 25% lower than the standard value, the system automatically enters a low tire pressure mode, which increases the weight of stability and reduces the weight of energy consumption. The maximum driving torque is limited according to the tire pressure decay ratio, and the slip ratio is stabilized in the safe range of 10% to 15% in combination with the road adhesion coefficient to prevent slippage and deviation, and reduce energy consumption.

[0035] Preferably, the power battery differential voltage diagnostic unit works in conjunction with the battery management system (BMS) to collect the voltage of each individual cell of the power battery in real time; the system calculates the difference between the highest and lowest voltages, and triggers a differential voltage alarm when the difference exceeds 100mV; for lithium iron phosphate batteries, passive resistance balancing and active capacitor balancing repair are performed; for ternary lithium batteries, a replacement command is directly output to avoid the risk of thermal runaway.

[0036] Preferably, the cloud platform management unit adopts a distributed storage architecture to receive and parse all data uploaded by the vehicle terminal; it visualizes information such as tire pressure, temperature, defects, torque, and battery differential pressure; when an anomaly is detected, it pushes alarm information via SMS and APP messages; the platform can store historical data and automatically generate statistical reports on tire life, energy consumption, and battery health to support intelligent operation and maintenance of the fleet. Example

[0037] This embodiment describes the workflow of Embodiment 1 above, as follows: Figure 1As shown, after the system is powered on and started, all functional units of the vehicle enter the initialization state. The tire pressure and temperature sensor, vehicle driving status sensor and power battery voltage acquisition module begin to collect signals in real time. The tire defect intelligent detection unit completes the algorithm model loading and enters the standby state. The tire pressure-drive coordination control unit and the power battery differential pressure diagnosis unit read the preset control parameters and fault judgment thresholds respectively. The vehicle terminal establishes a stable connection with the cloud platform management unit through the mobile communication network, and the entire system enters the continuous working mode.

[0038] The tire data remote monitoring unit collects real-time tire pressure and temperature information through tire pressure and temperature sensors installed on each wheel. The sensors wirelessly transmit the collected analog signals to the in-vehicle receiver, which converts them into digital signals conforming to the vehicle's communication standards via a CAN transceiver and then broadcasts them externally through the CAN bus. The on-board terminal continuously receives tire status information from the bus, encapsulates and processes the data according to a fixed period, and uploads it to the cloud platform management unit. When alarm signals such as abnormal tire pressure or excessive temperature are detected, the upload priority is immediately increased to achieve rapid reporting of fault information.

[0039] After receiving a tire X-ray image, the intelligent tire defect detection unit first performs image filtering, noise reduction, and size normalization. Then, it uses a lightweight residual convolutional network to extract multi-scale features, obtaining the tire's edge features, texture features, and high-level semantic features. The extracted feature maps are then fed into a dual-path polarized attention module for enhancement, highlighting defect-related features in the channel dimension and locating the defect area in the spatial dimension. Finally, the detection network identifies the defect type and locates the defect, and the detection results are sent to the vehicle terminal and vehicle controller via the CAN bus.

[0040] The tire pressure-drive coordination control unit acquires real-time information on tire pressure, tire temperature, vehicle speed, wheel speed, and road adhesion coefficient for each vehicle. Based on this information, it constructs a dual-objective optimization model for energy consumption and stability, and employs a tire pressure adaptive model predictive control algorithm to solve for the optimal drive torque distribution strategy. Under normal tire pressure conditions, the control unit prioritizes optimal energy consumption, rationally distributing drive torque between the front and rear axles and left and right wheels to improve overall vehicle fuel economy. When low tire pressure is detected in any wheel, the system automatically switches to a safety control mode. This mode uses a tire pressure-slip ratio fuzzy adaptive tuning algorithm to enhance stability weights, limits maximum output torque based on the degree of tire pressure decay, and keeps the tire slip ratio within a safe range to prevent dangerous situations such as vehicle slippage or veering.

[0041] The power battery differential voltage diagnostic unit interacts with the battery management system in real time, continuously acquiring voltage data of each individual cell within the power battery pack. It determines the battery pack's consistency status by calculating the voltage differences between individual cells. When the voltage difference exceeds a set threshold, the system identifies it as an abnormal battery voltage difference and triggers a fault alarm. Simultaneously, it executes differentiated processing strategies based on the battery type: initiating equalization and repair operations for lithium iron phosphate batteries and outputting replacement prompts for ternary lithium batteries. Related fault signals are simultaneously uploaded to the vehicle terminal and cloud platform.

[0042] The onboard terminal uniformly summarizes and formats tire status data, defect detection results, drive control information, and power battery diagnostic results. Regular status data is uploaded according to a set cycle, while abnormal fault data is pushed out in real-time with expedited updates. The cloud platform management unit parses, stores, and visualizes the received vehicle-side data, pushes multi-level alarms for abnormal information, and supports historical data queries, statistical analysis, and maintenance report generation. Management personnel can remotely monitor vehicle tire status, drive operation status, and battery health status through a cloud interface, achieving a closed loop for vehicle safety and maintenance management. Example

[0043] The only difference from Embodiment 1 above is that the intelligent tire defect detection unit uses a multi-scale residual attention lightweight detection algorithm, the process of which is as follows: The first step is to perform Gaussian filtering and scale normalization on the tire X-ray image to generate a 640×640 standard image; The second step involves four downsampling operations using a lightweight residual convolutional network to extract four levels of feature maps: shallow edges, mid-level textures, high-level semantics, and global contours. The third step is to input the four-level multi-scale feature map into the feature pyramid FPN module to complete the cross-layer weighted fusion of shallow edge features and high-level semantic features, thereby enhancing the multi-scale feature expression capability. The fourth step is to feed the fused feature map into the dual-path polarization attention module for feature enhancement; The dual-path polarization attention module is divided into a channel attention branch and a spatial attention branch. The channel attention branch performs global mean pooling and max pooling on the feature map, and after weighted fusion, it highlights the texture of defects and suppresses the background; the spatial attention branch performs position modeling on the features, locks the defect region, and strengthens the contour; the two outputs are multiplied element-wise to obtain the final enhanced feature map, which improves the detection rate of weak textures and small defects.

[0044] The fifth step involves using a lightweight YOLO inspection head combined with a dynamic focusing boundary loss function to complete bounding box regression. The focusing coefficient is dynamically adjusted based on the defect area, and the defect category, location, and confidence level are output.

[0045] This embodiment integrates the Feature Pyramid Network (FPN) module with a lightweight YOLO detection head into a multi-algorithm hybrid detection mechanism within a multi-scale residual attention algorithm, thereby further improving the accuracy, robustness, and real-time performance of tire defect detection. The FPN enables cross-layer weighted fusion of shallow edge features and high-level semantic features, enhancing the feature representation capability for defects of different sizes and effectively mitigating the problem of missed detection of weak textures and small defects. The lightweight YOLO detection head significantly reduces the number of model parameters and computational load while maintaining detection accuracy, making it more suitable for the computing power limitations of automotive-grade chips. The dynamic focusing boundary loss function, working synergistically with the lightweight YOLO detection head, adaptively adjusts weights based on the defect area, further improving defect localization accuracy and classification accuracy. The overall hybrid algorithm achieves efficient and stable detection of weak texture defects and small defects without increasing the hardware load on the vehicle, ensuring high reliability and practicality even under complex operating conditions.

[0046] Finally, it should be noted that the above embodiments are only used to illustrate the present invention and are not intended to limit the technical solutions described in the present invention. Therefore, although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the present invention. All technical solutions and improvements that do not depart from the spirit and scope of the present invention should be covered within the scope of the claims of the present invention.

Claims

1. A smart monitoring and collaborative control system for tires of new energy vehicles, characterized in that: It includes a tire data remote monitoring unit, a tire defect intelligent detection unit, a tire pressure-drive coordination control unit, a power battery differential pressure diagnosis unit, and a cloud platform management unit. Each unit is interconnected through the vehicle's CAN bus, and the data is uploaded to the cloud platform management unit via the vehicle terminal, realizing remote monitoring of tire status, intelligent defect identification, adaptive adjustment of drive torque, and linkage diagnosis of battery differential pressure faults.

2. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 1, characterized in that: The tire data remote monitoring unit includes a tire pressure and temperature sensor, a CAN transceiver, and an on-board remote monitoring terminal. The CAN transceiver is integrated into the tire monitoring display and converts tire pressure, temperature, and alarm signals into CAN message broadcasts. The on-board terminal uploads the data to the cloud platform at a fixed frequency.

3. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 2, characterized in that: The CAN message contains four types of signals: tire position, tire pressure, tire temperature, and alarm status. The message is sent once per second, and the vehicle terminal uploads the message to the cloud platform every 10 seconds.

4. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 1, characterized in that: The intelligent tire defect detection unit employs a multi-scale residual attention lightweight detection algorithm, which executes sequentially as follows: (1) Perform Gaussian noise filtering and scale normalization on the tire X-ray images to generate a standard 640×640 input image; (2) Using a lightweight residual convolutional network as the backbone, shallow edge features, mid-layer texture features, high-layer semantic features and global contour features are extracted sequentially through four layers of downsampling to form a four-level multi-scale feature map; (3) In the feature fusion stage, a dual-path polarization attention module is introduced to perform parallel weighted fusion of global mean pooling and max pooling on the channel dimension, and to perform adaptive localization enhancement of defect regions on the spatial dimension, and output the enhanced multi-scale fusion feature map. (4) The dynamic focusing boundary loss function is used to perform bounding box regression. The focusing coefficient is dynamically adjusted according to the defect area to complete the accurate detection, positioning and classification of three types of minor defects: open crown lines, sparse sidewall cords and impurities.

5. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 4, characterized in that: The lightweight residual convolutional network is composed of stacked multi-level depth-separable residual blocks. Each residual block contains three layers: 1×1 pointwise convolution, 3×3 depthwise convolution, and 1×1 pointwise convolution. During the training phase, an identity mapping residual branch is introduced for feature enhancement. During the inference phase, the multi-branch structure is fused into a single-path convolutional structure through structural reparameterization, reducing the computing power consumption of the vehicle terminal and enabling real-time inference.

6. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 4, characterized in that: The dual-path polarized attention module independently sets up a channel attention branch and a spatial attention branch. The channel branch is used to highlight the texture features of tire defects and suppress background noise, while the spatial branch is used to lock the geometric position of defects and enhance the boundary contour. The two outputs are multiplied element-wise to obtain the final attention-enhanced feature map.

7. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 1, characterized in that: The tire pressure-drive coordinated control unit incorporates a tire pressure adaptive model predictive control algorithm. This algorithm constructs a dual-objective optimization function of energy consumption and stability based on real-time tire pressure, tire temperature, vehicle speed, and wheel speed signals. Using tire slip ratio, drive motor efficiency, and yaw rate deviation as constraints, it solves the optimal drive torque distribution sequence for the next three control cycles through rolling time-domain optimization. Based on the degree of deviation of single tire pressure, it dynamically allocates the torque ratio of the left and right wheels / front and rear axles to achieve optimal synergy between energy consumption reduction and driving stability under low tire pressure conditions.

8. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 7, characterized in that: The tire pressure-drive coordinated control unit is equipped with a tire pressure-slip ratio fuzzy adaptive tuning algorithm; When a single tire pressure is detected to be 25% lower than the standard tire pressure, the low tire pressure mode is automatically triggered. The stability weight coefficient is increased and the energy consumption weight coefficient is decreased in real time through fuzzy rules, and the upper limit of the drive torque is linearly limited according to the tire pressure decay ratio. At the same time, the target value of the slip ratio is adaptively adjusted according to the real-time road surface adhesion coefficient, and the tire slip ratio is stably locked in the safe range of 10% to 15% to prevent slippage, deviation and excessive energy consumption.

9. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 1, characterized in that: The power battery differential voltage diagnostic unit collects the voltage of individual power battery cells in real time. When the voltage difference between individual cells exceeds the 100mV threshold, an alarm is triggered. For lithium iron phosphate batteries, equalization and charging repair are performed, and for ternary lithium batteries, direct replacement is performed.

10. The intelligent monitoring and collaborative control system for new energy vehicle tires according to claim 1, characterized in that: The cloud platform management unit adopts a distributed storage architecture and has functions such as data parsing, visualization, data anomaly verification, alarm information push and historical data statistical analysis.

Images