Vehicle body and chassis cooperative control system and method, vehicle and medium

Abstract

The invention relates to the field of vehicle control, and discloses a vehicle body and chassis cooperative control system and method, a vehicle and a medium, and the system comprises a data collection layer which is used for collecting chassis data and vehicle body data of the vehicle; the processing decision-making layer comprises a domain controller of a centralized architecture, and the domain controller is used for forwarding the chassis data to a chassis control model and forwarding the vehicle body data to a vehicle body control model; the fusion module is also used for carrying out weighted fusion on the first control instruction from the chassis control model and the second control instruction from the vehicle body control model to generate a target control instruction; and the instruction execution layer is used for performing cooperative control on a vehicle body and a chassis of the vehicle in response to the target control instruction. According to the invention, the complexity of data transmission is greatly reduced, and the cooperative control strategy is executed through the domain controller of the centralized architecture, so that the control cost is saved, and the controllability, stability and comfort of the vehicle are improved.

Description

Technical Field

[0001] The present application relates to the field of vehicle control, and in particular to a vehicle body chassis collaborative control system, method, vehicle and medium. Background Art

[0002] Traditional chassis and body control systems mostly adopt a distributed architecture. Each control unit, such as the electronic stability program ESP and the electronic control suspension ECS, operates independently and exchanges data through the CAN bus (controller area network).

[0003] However, in related technologies, this distributed architecture lacks deep data sharing and coordination between control units. For example, the brake and suspension systems cannot coordinate and adjust in real time during emergency braking. This lack of close linkage between brake pressure regulation and vehicle posture stability, and between steering angle adjustment and suspension stiffness adjustment, severely impacts the vehicle's overall stability and safety. Furthermore, control strategies often rely on pre-set, fixed rules, making it difficult to dynamically optimize based on the vehicle's real-time state, which can easily affect overall vehicle control accuracy. Summary of the Invention

[0004] The embodiments of the present application provide a body-chassis collaborative control system, method, vehicle, and medium to solve the problem in related technologies that the body-chassis cannot be centrally and collaboratively controlled, affecting overall safety and stability.

[0005] An embodiment of the present application provides a body-chassis collaborative control system, including: a data acquisition layer, used to collect chassis data and body data of a vehicle; a processing decision layer, including a domain controller with a centralized architecture, the domain controller is used to forward the chassis data to a chassis control model and forward the body data to a body control model; and is also used to weightedly fuse a first control instruction from the chassis control model with a second control instruction from the body control model to generate a target control instruction; an instruction execution layer, used to collaboratively control the body and chassis of the vehicle in response to the target control instruction.

[0006] In one embodiment of the present application, the processing decision layer further includes: a chassis control model for preprocessing the chassis data, inputting the preprocessed chassis data into a preset multi-layer convolutional neural network model for feature extraction, and outputting a first feature vector; performing similarity matching on the first feature vector and a preset chassis control strategy to determine a first control strategy; using a deep learning algorithm to analyze the state and change trend of the chassis based on time series characteristics, and generating a first control instruction for adjusting the chassis in combination with the first control strategy, wherein the chassis data includes at least one of wheel speed, suspension displacement, steering angle, and brake pressure; a body control model for preprocessing the body data, inputting the preprocessed body data into a preset recurrent neural network model for pattern recognition, and outputting a body motion pattern; performing similarity matching on the body motion pattern and a preset body control strategy to determine a second control strategy; and generating a second control instruction for adjusting the body based on the second control strategy and a preset deep learning algorithm; the body data includes at least one of body posture angle, lateral acceleration, and longitudinal acceleration.

[0007] In one embodiment of the present application, the domain controller performs weighted calculation on the first control instruction of the chassis and the second control instruction of the body according to a preset chassis weight and a preset body weight, respectively, and generates a target control instruction to be sent to an actuator group, wherein the actuator group includes at least one of a braking actuator, a steering actuator, a suspension adjustment actuator, and a body stability actuator.

[0008] In one embodiment of the present application, the domain controller performs hierarchical caching on the chassis data and the body data according to a hierarchical caching strategy; the domain controller determines the sending order of each data in the data queue to be transmitted according to the data priority, and the chassis data and the body data use different data queues and transmit data respectively through different direct memory access channels; wherein, the hierarchical caching strategy uses multiple memories for caching according to the data priority of the received data, each memory has different read and write speeds and access delays, and the data priority is determined by the data type or data risk assessment, and the priority of the chassis data is higher than the priority of the body data.

[0009] In one embodiment of the present application, the domain controller dynamically allocates the CAN bus bandwidth required for data transmission for chassis control, body control and fault diagnosis control according to the token bucket algorithm; wherein, the token bucket algorithm controls the traffic by the token generation rate and the number of tokens, and the token bucket algorithm satisfies at least one of the following conditions: the token generation rate increases with the weight coefficient of the data type, and the data of each data type is configured with different weight coefficients according to the data priority; the token bucket algorithm includes a dual-bucket hierarchical token mechanism of high-priority bucket tokens and low-priority bucket tokens, the high-priority bucket token is configured with a token number capacity for data with high priority, and the low-priority bucket token is configured with a token number capacity for data with low priority, and the token generation rate corresponding to the high-priority bucket token is greater than the token generation rate of the low-priority bucket token; the token bucket algorithm dynamically adjusts the refresh period according to the queue length of the data queue; the weight coefficient of the data is positively correlated with the urgency, and the number of tokens is allocated to the data based on the weight coefficient and the urgency.

[0010] In one embodiment of the present application, the domain controller further includes: detecting the load of instantaneous traffic through a preset smoothing factor: if it is detected that any load transmitted to the chassis control model and the body control model is greater than the preset load, overload is determined, and the operation frequency of the overloaded data is reduced based on the data priority.

[0011] In one embodiment of the present application, the processing decision layer also includes: a fault diagnosis module, which is used to monitor the operating status of the target object in the vehicle and determine whether the target object is faulty based on the operating status; if it is detected that the target object has a fault, a fault alarm message is generated, and the backup control scheme corresponding to the faulty object is switched to respond; if it is detected that the target object has not a fault, monitoring is continued, and the target object includes at least one of the sensor group in the data acquisition layer, the actuator group in the instruction execution layer, the chassis control model, and the body control model.

[0012] In one embodiment of the present application, it also includes: an optimization update layer, which is used to associate and map the received vehicle actual feedback data with the target control instruction to form an optimized data set, and optimize the chassis control model and the body control model based on the optimized data set to update the internal parameters of the chassis control model and the body control model. The vehicle actual feedback data is actual data generated after the actuator group executes the target control instruction.

[0013] An embodiment of the present application also provides a body-chassis collaborative control method, including: collecting chassis data and body data of a vehicle; using a domain controller to receive the chassis data and the body data, forwarding the chassis data to a chassis control model and forwarding the body data to a body control model; weightedly fusing a first control instruction from the chassis control model with a second control instruction from the body control model to generate a target control instruction; and in response to the target control instruction, collaboratively controlling the body and chassis of the vehicle.

[0014] An embodiment of the present application also provides a vehicle that adopts the system of any of the above embodiments.

[0015] An embodiment of the present application further provides a computer-readable storage medium, which stores a computer program. When the computer program is executed by a processor, the method of any of the above embodiments is implemented.

[0016] In the solution implemented by the body-chassis collaborative control system, method, vehicle, and medium provided above, chassis data and body data are forwarded through a domain controller, and the first control instruction from the chassis control model and the second control instruction from the body control model are weightedly fused to generate a target control instruction. This not only greatly reduces the complexity of data transmission, but also provides timely and accurate data support for the chassis control model and the body control model, ensuring that each model can quickly respond to changes in the vehicle's operating state and generate accurate target control instructions, thereby achieving efficient collaborative control of the chassis and body, and comprehensively improving the vehicle's handling performance and driving safety; in response to the target control instruction, the body and chassis of the vehicle are collaboratively controlled. On the one hand, the chassis control model and the body control model work together to ensure that the vehicle can maintain good chassis buffering performance and a stable body posture when driving; on the other hand, the execution of the collaborative control strategy by the centralized domain controller not only saves control costs, but also effectively improves the vehicle's handling, stability, and comfort, meeting the needs of users in different driving scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The accompanying drawings are incorporated into and constitute a part of the specification, illustrate embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be derived from these drawings without inventive effort.

[0018] In the attached figure: Figure 1 A schematic diagram of an exemplary architecture of a vehicle-chassis collaborative control system that can be applied in an embodiment of the present application; Figure 2 A schematic structural diagram of a vehicle body chassis collaborative control system provided in an embodiment of the present application; Figure 3 A schematic diagram of the principle of the vehicle body chassis collaborative control system provided in an embodiment of the present application; Figure 4 A data partition state transition diagram in the vehicle body chassis collaborative control system provided in an embodiment of the present application; Figure 5 A schematic diagram of the token bucket algorithm structure in the vehicle body chassis collaborative control system provided in an embodiment of the present application; Figure 6 A schematic flow chart of a vehicle body chassis collaborative control method provided in an embodiment of the present application; Figure 7 A schematic structural diagram of an electronic device in one embodiment of the present application; Figure 8 Another structural diagram of an electronic device in an embodiment of the present application. DETAILED DESCRIPTION

[0019] The following describes the embodiments of the present application through specific examples. Those skilled in the art can easily understand other advantages and functions of the present application from the contents disclosed in this specification. The present application can also be implemented or applied through other different specific embodiments. The details in this specification can also be modified or changed based on different viewpoints and applications without departing from the spirit of the present application. The following embodiments and features in the embodiments can be combined with each other without conflict.

[0020] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application and are not drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component can be changed at will, and the component layout type may also be more complicated.

[0021] In the following description, a large number of details are discussed to provide a more thorough explanation of the embodiments of the present application. However, it is obvious to those skilled in the art that the embodiments of the present application can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present application difficult to understand.

[0022] Figure 1An exemplary architectural diagram of a body-chassis collaborative control system that can be applied in an embodiment of the present application is provided. Vehicle 100 includes, but is not limited to, new energy vehicles such as pure electric vehicles, extended-range electric vehicles, plug-in hybrid vehicles, hydrogen energy vehicles, and can also be internal combustion engine vehicles, which will not be repeated here. Among them, the vehicle 100 may include a perception system, a display device, and a computing platform, wherein the perception system may include several sensors for sensing information about the environment surrounding the vehicle 100. For example, the perception system may include a positioning system, and the positioning system may be a global positioning system (GPS), or a Beidou system or other positioning systems, an inertial measurement unit (IMU), a lidar, a millimeter-wave radar, an ultrasonic radar, and a camera device. One or more of the body-chassis collaborative control system 101 is built into the vehicle 100.

[0023] Some or all functions of the vehicle 100 may be controlled by a computing platform. The computing platform may include multiple processors. A processor is a circuit with signal processing capabilities. In one implementation, the processor may be a circuit with instruction reading and execution capabilities, such as a central processing unit (CPU), a microprocessor, a graphics processing unit (GPU) (which can be understood as a microprocessor), or a digital signal processor (DSP). In another implementation, the processor may implement certain functions through the logical relationship of a hardware circuit. The logical relationship of the hardware circuit is fixed or reconfigurable. For example, the processor is a hardware circuit implemented by an application-specific integrated circuit (ASIC) or a programmable logic device (PLD), such as an FPGA. In a reconfigurable hardware circuit, the process of the processor loading a configuration document to implement the hardware circuit configuration can be understood as the process of the processor loading instructions to implement the functions of some or all of the above units. In addition, it can also be a hardware circuit designed for artificial intelligence, which can be understood as an ASIC, such as a neural network processing unit (NPU), a tensor processing unit (TPU), a deep learning processing unit (DPU), etc. In addition, the computing platform can also include a memory, which is used to store instructions. Some or all of the multiple processors can call the instructions in the memory to implement corresponding functions.

[0024] In related technologies, with the development of intelligent and electrified vehicles, coordinated control between vehicle subsystems is becoming increasingly important. In traditional vehicle control systems, chassis and body control are often independent of each other, and cannot be comprehensively optimized based on the overall vehicle state. For example, when cornering at high speeds, the chassis suspension system and the body stability system cannot work in coordination, resulting in severe vehicle roll, reduced controllability, and even affecting driving safety. Therefore, a system that can integrate chassis and body data and achieve coordinated control is needed to improve vehicle performance and safety under various operating conditions.

[0025] In view of this, the embodiments of the present application provide a vehicle body chassis collaborative control system, method, vehicle and medium, please refer to Figure 2 , Figure 2A schematic structural diagram of a vehicle body chassis collaborative control system provided in an embodiment of the present application is described in detail as follows: The data collection layer 21 is used to collect chassis data and body data of the vehicle; The processing decision layer 22 includes a centralized domain controller 220, which is used to forward chassis data to the chassis control model and forward body data to the body control model; and is also used to perform weighted fusion of a first control instruction from the chassis control model and a second control instruction from the body control model to generate a target control instruction; The instruction execution layer 230 is used to coordinately control the vehicle body and chassis in response to the target control instruction.

[0026] For example, various sensors are used to collect vehicle body and chassis data. For example, consider wheel speed sensors. Electromagnetic wheel speed sensors utilize the principle of electromagnetic induction. When the wheel rotates, the induction coil inside the sensor interacts with the wheel gear ring, generating an alternating induced electromotive force (EMF). The frequency of the induced electromotive force is proportional to the wheel speed. The wheel speed can be determined by measuring the frequency of the induced electromotive force. For example, Hall effect wheel speed sensors utilize the Hall effect. When the magnetic field changes, the Hall element generates a voltage signal related to the magnetic field strength. The wheel speed is determined by detecting changes in this voltage signal. Displacement sensors utilize the principle of electromagnetic coupling. When the iron core moves between the primary and secondary coils, the induced voltage in the secondary coil changes. This voltage change is measured to calculate suspension displacement. Acceleration sensors measure acceleration by detecting changes in capacitance, resistance, or inductance based on the balance between the inertial force generated by a mass under acceleration and the elastic force of an elastic element. Gyroscopes utilize the principle of conservation of angular momentum to measure the angular velocity of an object by detecting the precession of a rotating component. It collects chassis data (wheel speed, steering angle) and body data (attitude angle, lateral acceleration) in real time, converts the physical quantities of various sensors into electrical signals, amplifies and filters them through conditioning circuits, and then converts them into digital signals. The signals are then transmitted to the domain controller via the CAN bus, enabling all-round collection of multi-source data.

[0027] The chassis control model and body control model are constructed based on vehicle dynamics principles. For example, the chassis control model's dynamics model takes into account factors such as vehicle mass, wheel-ground friction, and the elasticity and damping characteristics of the suspension. When calculating braking force distribution, Newton's second law is applied. Based on information such as vehicle acceleration and wheel speed, mathematical formulas are used to calculate the braking force required for each wheel to maintain stable braking. The body control model is also based on vehicle dynamics. Incorporating parameters such as the vehicle's inertia and center of mass position, dynamic equations are used to describe the vehicle's motion under various external forces, such as roll, pitch, and yaw.

[0028] Sensor data is converted into control commands to achieve precise control of the vehicle chassis and body. In vehicle control systems, different control commands have varying impacts on vehicle performance under different operating conditions. Therefore, these commands need to be balanced through a weighted approach, typically based on actual vehicle test data, simulation results, and empirical knowledge. For example, extensive testing of high-speed driving conditions may determine that chassis control commands contribute significantly to vehicle stability under these conditions, thus assigning them a higher weight. Weighted fusion can employ various mathematical methods, such as a simple weighted average, where two control commands are multiplied by their respective weights and then added together. Alternatively, more complex weighted fusion methods based on fuzzy logic can be employed. Based on various vehicle operating parameters, such as speed, steering angle, and road adhesion coefficient, a fuzzy inference system dynamically adjusts the weights, achieving more intelligent and precise control command fusion.

[0029] Specifically, the data acquisition layer obtains the vehicle's chassis data and body data through various sensors. For chassis data, sensors are installed in key parts of the chassis, such as wheels, suspension, and steering systems. Displacement sensors and acceleration sensors are set in the suspension system. The displacement sensors can measure the compression and extension of the suspension, and the acceleration sensors can sense the vibration of the suspension, thereby obtaining the dynamic posture information of the chassis.

[0030] The processing decision layer is a centralized domain controller. After receiving chassis and body data from the data acquisition layer, the domain controller first preprocesses the data, including but not limited to filtering and removing sensor noise and interference signals. For example, a Kalman filter algorithm uses the state estimate at the previous moment and the measured value at the current moment to obtain a more accurate state estimate through recursive calculation, thereby improving data accuracy and stability. Secondly, the preprocessed data is analyzed to convert the digital signals transmitted by the sensors into physical quantities. The domain controller then forwards the preprocessed chassis data to the chassis control model and the body data to the body control model. The chassis control model then uses the corresponding control algorithm based on the chassis data to calculate a first control instruction. For example, when the vehicle brakes, the chassis control model determines whether the wheels are about to lock based on the wheel speed sensor data. If it detects a tendency for the wheels to lock, it calculates the appropriate braking force distribution and generates a first control instruction to adjust the braking force on each wheel to prevent wheel lock and ensure vehicle braking stability. The body control model generates a second control instruction based on the body data. For example, when the vehicle turns, the body control model calculates the vehicle's roll angle and yaw rate based on the body posture sensor data. Combined with the vehicle dynamics model, it generates a second control instruction to adjust the output of the body stability system and control the stiffness of the anti-roll bar to reduce body roll. Finally, the domain controller performs a weighted fusion of the first control command from the chassis control model and the second control command from the body control model. Based on the vehicle's current driving conditions (such as high-speed driving, low-speed cornering, and emergency braking), and pre-set weighting rules, the two control commands are assigned different weights. For example, at high speeds, chassis stability is crucial to vehicle safety, so the first control command from the chassis control model is given a higher weight. Meanwhile, at low-speed cornering, vehicle posture control is more critical, so the second control command from the body control model is appropriately weighted. Through this weighted summation, a target control command is generated, which comprehensively considers the control requirements of both the chassis and the body to achieve optimal overall vehicle control.

[0031] The command execution layer receives target control commands from the decision-making layer and translates them into actual actions to coordinately control the vehicle's body and chassis. Regarding chassis control, when a target control command involves adjusting suspension stiffness, the command execution layer controls the solenoid valves or motors in the suspension system, altering the oil flow or spring preload within the suspension to adjust the stiffness. For example, at high speeds, to improve vehicle handling stability, if the target control command requires increasing suspension stiffness, the command execution layer controls the solenoid valves to reduce the suspension oil flow area, stiffening the suspension. Regarding the chassis' braking system, if a target control command requires increasing braking force on a specific wheel, the command execution layer controls the pressure regulator in the brake line to increase the oil pressure in the brake cylinder at that wheel, thereby increasing friction between the brake pads and discs and achieving increased braking force. Regarding body control, when a target control command requires adjusting vehicle posture, the command execution layer activates the corresponding actuators. For example, when the vehicle is turning, if the target control instruction requires reducing the body roll, the instruction execution layer controls the electric anti-roll bar in the body stabilization system, drives the anti-roll bar to twist through the motor, increases its anti-roll ability, and reduces the body roll angle.

[0032] For example, under complex road conditions, the chassis control model and the body control model work together to ensure both good chassis cushioning and a stable body position when navigating bumpy roads. At high speeds, chassis control is prioritized to ensure vehicle stability; in urban traffic, body control is prioritized to enhance ride comfort. This intelligent, coordinated control strategy effectively enhances vehicle handling, stability, and comfort, meeting user needs in diverse driving scenarios. The command execution layer accurately and efficiently executes target control commands, providing a strong foundation for coordinated vehicle control. Precise suspension stiffness adjustment and braking force distribution allow the vehicle to better adapt to varying road conditions and driving requirements, enhancing driving safety and handling stability. For example, in emergency evasive maneuvers, the fast-responding braking system and optimal suspension adjustments ensure rapid vehicle deceleration and stable position, preventing dangerous situations such as rollover. Precise control of body position, such as effectively reducing body roll during cornering, enhances passenger comfort.

[0033] Through the above methods, the data collection layer can obtain various information of the vehicle chassis and body in real time and accurately. The centralized domain controller, which handles the decision-making layer, performs a weighted fusion of the primary and secondary control commands, enabling accurate control decisions based on the vehicle's real-time state. This weighted fusion mechanism enables the vehicle to flexibly adjust the emphasis of chassis and body control according to different driving conditions, optimizing overall performance. The chassis and body control models work together to ensure both good chassis cushioning and a stable body posture during driving. Furthermore, the centralized domain controller executes collaborative control strategies, saving control costs while effectively improving vehicle handling, stability, and comfort, meeting user needs in diverse driving scenarios.

[0034] In related technologies, traditional models are not accurate enough in extracting features from chassis and body data, making it difficult to capture the complex correlations hidden in the data, resulting in inaccurate matching of control strategies. At the same time, they lack sufficient intelligence and adaptability in analyzing chassis status and changing trends, and identifying body motion patterns, and are unable to dynamically adjust control instructions based on real-time data. This results in poor control of the chassis and body in complex road conditions and dynamic driving, affecting the vehicle's handling, stability, and safety.

[0035] In some embodiments, the processing decision layer further includes: a chassis control model for preprocessing chassis data, inputting the preprocessed chassis data into a preset multi-layer convolutional neural network model for feature extraction, and outputting a first feature vector; performing similarity matching between the first feature vector and a preset chassis control strategy to determine a first control strategy; employing a deep learning algorithm to analyze a chassis state and change trend based on time series characteristics, and generating a first control instruction for adjusting the chassis in combination with the first control strategy, wherein the chassis data includes at least one of wheel speed, suspension displacement, steering angle, and brake pressure; The body control model is used to preprocess the body data, input the preprocessed body data into a preset recurrent neural model for pattern recognition, and output the body motion pattern; the body motion pattern is matched with the preset body control strategy for similarity to determine the second control strategy; according to the second control strategy and the preset deep learning algorithm, a second control instruction for adjusting the body is generated; the body data includes at least one of the body posture angle, lateral acceleration, and longitudinal acceleration.

[0036] Exemplarily, the chassis control model receives chassis data, which includes at least one of wheel speed, suspension displacement, steering angle, and brake pressure. The chassis data is preprocessed, which includes data cleaning, removing outliers and noise, and converting chassis data of different magnitudes into a unified numerical range, such as [0,1], through data normalization. For example, the wheel speed is converted from 0-3000r / min to a value of 0-1 for subsequent model processing; data is completed, and interpolation methods such as linear interpolation and polynomial interpolation are used to supplement missing data points to ensure data integrity and continuity. The preprocessed chassis data is fed into a pre-built multi-layer convolutional neural network (CNN) model. The CNN model consists of multiple convolutional layers, pooling layers, and activation function layers. The convolutional layers convolve the input data with kernels of varying sizes to extract local features. For example, for wheel speed data that varies over time, the kernels can capture both sudden and steady changes in speed. Pooling layers, such as max pooling and average pooling, downsample the feature maps output by the convolutional layers to reduce feature dimensionality while retaining important features. Activation functions, such as ReLU, introduce nonlinearity, enabling the model to fit complex nonlinear relationships. After multiple layers of processing, the CNN model outputs a first feature vector, which contains the key features of the chassis data. This first feature vector is then matched against the strategy feature vectors in a pre-built chassis control strategy library. This library, constructed based on extensive historical data and expert experience, contains optimal control strategies for various operating conditions, such as suspension control during rapid acceleration and braking control during sharp cornering. Similarity matching can be performed using methods such as cosine similarity and Euclidean distance to calculate the similarity between the first eigenvector and the eigenvectors of each strategy. The strategy with the highest similarity is then selected as the first control strategy. Deep learning algorithms (such as LSTM (Long Short-Term Memory)) are used to analyze the chassis' status and changing trends based on time series characteristics. LSTM models can process time series data and capture long-term data dependencies. For example, by analyzing suspension displacement changes over a period of time, the suspension's state can be predicted within the next few seconds. Combined with the first control strategy, the LSTM model analyzes the chassis's status and trends and generates specific first control instructions, such as adjusting suspension stiffness parameters, changing brake pressure, and correcting steering angle, to achieve precise chassis adjustment.

[0037] Exemplarily, the body control model receives vehicle body data, including at least one of body attitude angle, lateral acceleration, and longitudinal acceleration. Preprocessing methods are similar to those used in the chassis control model, including data cleaning to remove abnormal data caused by sensor drift or external interference; data normalization to convert the vehicle body data to a uniform scale, for example, converting lateral acceleration from -5 to 5 meters per second squared to values ​​from -1 to 1; and data smoothing, using methods such as sliding averages to reduce high-frequency noise and smooth the data curve, facilitating subsequent pattern recognition. The preprocessed vehicle body data is then input into a pre-set recurrent neural network model (such as an LSTM or GRU gated recurrent unit). Recurrent neural networks have memory capabilities and can process sequential data, making them suitable for identifying patterns in vehicle body data over time. By learning from the input vehicle body data sequence, the recurrent neural model identifies vehicle body motion patterns, such as straight-line driving mode, turning mode (left turn, right turn), braking mode, and acceleration mode. For example, when lateral acceleration is zero and longitudinal acceleration is stable, the model identifies a straight-line driving mode. When lateral acceleration is positive and the vehicle's posture angle deflects to the left, it identifies a left-turning mode. The recurrent neural network model outputs an identifier or feature vector for the vehicle's motion mode. The vehicle's motion mode is then matched against a pre-defined control strategy library, which contains optimal control strategies for different vehicle motion modes, such as vehicle stability control in cornering mode and vehicle posture adjustment in braking mode. The strategy with the highest matching score is selected as the second control strategy. Based on the second control strategy and a pre-defined deep learning algorithm (such as a deep belief network (DBN)), a second control command for adjusting the vehicle's body is generated. The DBN, composed of a stack of restricted Boltzmann machines (RBMs), extracts deep features from data through unsupervised learning and combines it with supervised learning for prediction. Combined with the second control strategy, the DBN conducts in-depth analysis of the vehicle's motion mode and generates specific second control commands, such as adjusting parameters of the vehicle stability system or controlling the force of the anti-roll bar, to achieve effective vehicle adjustment.

[0038] Through this approach, the chassis control model utilizes a multi-layer CNN to extract key features of chassis data, combined with an LSTM to analyze state trends and generate control instructions, improving the adaptability and foresight of chassis control. The body control model uses a recurrent neural network model to identify body motion patterns and, combined with a DBN to generate control instructions, enhances the accuracy and effectiveness of body control. This embodiment addresses the shortcomings of traditional control models in processing complex dynamic data, better capturing hidden features and temporal dependencies in the data, enabling more precise control strategy selection and optimized control instruction generation.

[0039] In the related art, in a vehicle control system, since the first control instruction of the chassis and the second control instruction of the vehicle body are often generated based on their respective control models, there may be problems of instruction conflict or insufficient coordination.

[0040] To solve the above problem, in some embodiments, the domain controller performs weighted calculation on the first control instruction of the chassis and the second control instruction of the body according to the preset chassis weight and the preset body weight, respectively, and generates a target control instruction sent to the actuator group. The actuator group includes at least one of a braking actuator, a steering actuator, a suspension adjustment actuator and a body stability actuator.

[0041] For example, the domain controller pre-stores preset chassis and body weights. For example, during high-speed straight-line driving, vehicle stability relies on chassis control, so the preset chassis weight is set to 0.7 and the preset body weight is 0.3. During low-speed cornering, however, body posture control becomes more important, so the preset chassis weight is adjusted to 0.4 and the preset body weight is adjusted to 0.6. The domain controller receives a first control instruction from the chassis control model and a second control instruction from the body control model. For example, the first control instruction may include brake pressure adjustment values, steering angle correction values, and suspension stiffness parameters; the second control instruction may include the intervention intensity of the vehicle stability system and the adjustment force of the anti-roll bar. The domain controller performs a weighted calculation on the first control instruction based on the preset chassis weights, multiplying each parameter of the first control instruction by the preset chassis weights. Simultaneously, the domain controller performs a weighted calculation on the second control instruction based on the preset body weights, multiplying each parameter of the second control instruction by the preset body weights.

[0042] For example, if the brake pressure adjustment value in the first control command is 500kPa and the preset chassis weight is 0.6, the weighted brake pressure adjustment value is 500 × 0.6 = 300kPa. If the body stability intervention intensity in the second control command is 80% and the preset body weight is 0.4, the weighted intervention intensity is 80% × 0.4 = 32%. The weighted first and second control commands are fused. For control parameters of the same type, the integration is performed by summing or averaging. For control parameters of different types, they are directly combined to form a target control command. The target control command contains specific control requirements for each actuator, such as the target pressure for the brake actuator and the target angle for the steering actuator.

[0043] For example, the domain controller parses the generated target control instructions according to the actuator type, extracting the control parameters corresponding to each actuator. The domain controller then transmits the parsed control parameters to the corresponding actuator via the vehicle's internal communication network. After receiving the control parameters, each actuator converts them into specific actions according to a pre-set control algorithm. For example, a brake actuator adjusts the hydraulic or pneumatic system to achieve a target brake pressure; a steering actuator drives the steering mechanism through a motor to achieve a target steering angle; a suspension adjustment actuator adjusts the hydraulic valve or motor to change the suspension stiffness; and a body stability actuator achieves body stability control by braking a single wheel or adjusting power distribution.

[0044] This approach uses a domain controller to perform a weighted fusion of chassis and body control commands, then sends the target control command to the actuator group. This effectively addresses the lack of coordination between chassis and body commands in traditional vehicle control. Sending the command to the actuator group ensures accurate transmission and effective execution of the control command, enabling the vehicle to demonstrate excellent performance in various operating conditions. For example, during an emergency avoidance maneuver, the target control command coordinates the braking, steering, and body stability actuators, enabling rapid vehicle deceleration and smooth steering while maintaining vehicle stability, significantly improving driving safety.

[0045] In related technologies, the transmission and caching of chassis and body data in vehicle control systems face numerous challenges. Vehicle operation generates a large amount of real-time data, and improper caching and transmission can easily lead to data congestion and latency.

[0046] To solve the above problems, in some embodiments, the domain controller caches chassis data and body data in a hierarchical manner according to a hierarchical caching strategy; the domain controller determines the sending order of each data in the data queue to be transmitted according to the data priority, and the chassis data and the body data use different data queues and transmit the data respectively through different direct memory access channels; wherein, the hierarchical caching strategy uses multiple memories for caching according to the data priority of the received data, each memory has different read and write speeds and access delays, and the data priority is determined by the data type or data risk assessment, and the priority of the chassis data is higher than that of the body data.

[0047] For example, the domain controller receives chassis data such as wheel speed, suspension displacement, steering angle, and brake pressure. It also receives body data such as body attitude angle, lateral acceleration, longitudinal acceleration, and interior temperature. First, the data is prioritized based on data type or risk assessment, with chassis data taking precedence over body data. Within chassis data, data directly related to vehicle safety, such as brake pressure and wheel speed, has a higher priority. Within body data, data that impacts vehicle stability, such as body attitude angle and lateral acceleration, has a relatively higher priority, while data such as interior temperature has a lower priority. A hierarchical caching strategy uses multiple memory types based on data priority for caching, each with varying read / write speeds and access latencies. Typically, high-priority data is cached in memory with fast read / write speeds and low access latency, such as static random access memory (SRAM); medium-priority data can be stored in dynamic random access memory (DRAM); and low-priority data can be stored in memory with slower read / write speeds but larger storage capacities, such as flash memory. For example, the brake pressure data in the chassis data has the highest priority and is stored in SRAM; the body posture angle data in the body data has the second highest priority and is stored in DRAM; low-priority data such as the interior temperature is stored in Flash; the domain controller stores the chassis data and body data with determined priorities in the corresponding memories respectively.

[0048] During the caching process, the memory is managed. For example, when there is insufficient cache space for high-priority data, preset rules are followed, such as eliminating low-priority data to release space, to ensure that high-priority data can be cached in a timely manner. At the same time, the cached data is marked to indicate its priority and data type to facilitate data processing and transmission.

[0049] For example, the domain controller reads chassis and body data to be transmitted from a cache and constructs separate data queues based on data type: a chassis data queue and a body data queue. Within each queue, data is sorted according to priority, with higher-priority data placed at the front of the queue and sent first. For example, in the chassis data queue, brake pressure data is placed first, followed by wheel speed data; in the body data queue, body attitude angle data is placed first, followed by interior temperature data. Because chassis data has a higher priority than body data, the chassis data queue is sent before the body data queue in the overall sending order. When both queues have data to send, one approach is to send the data in the chassis data queue first, followed by the data in the body data queue. Another approach is to dynamically allocate bandwidth to the two queues based on weight, priority, or urgency for separate transmission. Within the same queue, data is sent in strict order, following the sort order of the queues. The domain controller allocates different DMA channels (direct memory access channels) for the chassis data queue and the body data queue. A DMA channel is a channel that enables data transfer between memory and peripherals without CPU intervention. Different DMA channels are independent of each other and can transmit data in parallel. Typically, DMA channels with higher bandwidth and faster transmission speeds are allocated to high-priority chassis data queues to ensure that their data can be transmitted quickly. Through their respective DMA channels, the data in the chassis data queue and the body data queue are transmitted to the corresponding models respectively. During the transmission process, the domain controller monitors the data transmission status. If a transmission error or congestion occurs, timely measures such as retransmission and adjustment of the transmission rate are taken to ensure accurate and timely data transmission.

[0050] Through the above methods, through hierarchical caching strategies and differentiated data transmission mechanisms, the efficiency and priority issues in vehicle data transmission and caching are effectively solved; among them, the hierarchical cache reasonably allocates storage resources according to data priority, ensuring that high-priority chassis data can be cached quickly, while optimizing the utilization of storage resources and reducing system costs.

[0051] By building different data queues and using different DMA channels to transmit data differentially, priority and parallel transmission of high-priority data are guaranteed, the efficiency and orderliness of data transmission are improved, and the transmission delay of key data is reduced. This enables the processing decision-making layer to obtain important chassis and body data in a timely manner, quickly generate control instructions, and improve the timeliness and accuracy of vehicle control.

[0052] In vehicle control systems, the CAN bus, a key channel for data transmission, has limited bandwidth resources. Systems such as chassis control, body control, and fault diagnosis generate large amounts of data. Improper bandwidth allocation can lead to the following issues: High-priority data, such as chassis emergency braking signals and fault information, can be delayed due to bandwidth being occupied by lower-priority data, impacting vehicle safety. Furthermore, fixed bandwidth allocation cannot adapt to the dynamic changes in data transmission requirements across different systems, leading to wasted or insufficient bandwidth resources.

[0053] To solve the above problem, in some embodiments, the domain controller dynamically allocates the CAN bus bandwidth required for data transmission for chassis control, body control, and fault diagnosis control according to the token bucket algorithm; wherein the token bucket algorithm controls the flow by the token generation rate and the number of tokens, see Figure 5 , the token bucket algorithm meets at least one of the following conditions: The token generation rate increases with the weight coefficient of the data type. Each data type has a different weight coefficient depending on the data priority. The token bucket algorithm includes a dual-bucket hierarchical token mechanism consisting of high-priority bucket tokens and low-priority bucket tokens. The high-priority bucket tokens are used to configure the token capacity for high-priority data, while the low-priority bucket tokens are used to configure the token capacity for low-priority data. The token generation rate corresponding to the high-priority bucket tokens is greater than the token generation rate of the low-priority bucket tokens. The token bucket algorithm dynamically adjusts the refresh period according to the queue length of the data queue; The weight coefficient of the data is positively correlated with the urgency, and the number of tokens is allocated to the data based on the weight coefficient and the urgency.

[0054] Exemplarily, the domain controller pre-configures key parameters of the token bucket algorithm, including weight coefficients for different data types (e.g., weight coefficients that are positively correlated with data priority and urgency, with higher priority and greater urgency resulting in larger weight coefficients), the token capacity of the high-priority and low-priority buckets, and the initial token generation rate. For example, the weight coefficient for brake pressure data in chassis control is set to 0.8, for in-vehicle temperature data in body control to 0.3, and for fatal fault data in fault diagnosis to 0.9. The high-priority bucket capacity is set to 100 tokens, and the low-priority bucket to 50 tokens. The initial token generation rate for the high-priority bucket is 20 tokens / ms, while that for the low-priority bucket is 5 / ms.

[0055] For example, the domain controller receives data from chassis control, body control, and fault diagnosis control, categorizes it according to data type and priority, and places high-priority data, such as chassis emergency braking signals and critical fault information, into a high-priority bucket, while low-priority data, such as body comfort adjustment data, into a low-priority bucket. The domain controller also records the queue lengths of each data queue. A token bucket algorithm generates tokens at a set token generation rate (which increases with the data type weight coefficient, with higher weight coefficients corresponding to higher generation rates). The high-priority bucket generates tokens at a higher rate than the low-priority bucket. For example, when transmitting high-priority chassis data, the high-priority bucket's token generation rate increases to 25 tokens / ms due to its high weight coefficient. When transmitting low-priority body data, the low-priority bucket maintains a token generation rate of 5 / ms. Furthermore, the refresh cycle is dynamically adjusted based on the length of the data queue. Longer queue lengths result in shorter refresh cycles, accelerating token generation and allocation. Data transmission consumes tokens, with each data unit assigned a certain number of tokens (based on the weight coefficient and urgency, with higher urgency units consuming more tokens to prioritize bandwidth).

[0056] For example, the domain controller allocates CAN bus bandwidth based on the number of tokens held by each system data. A higher number of tokens grants more bandwidth for data transmission. For example, if fault diagnosis holds 10 tokens, chassis control holds 60, and body control holds 30, chassis control will be prioritized for bandwidth. Data transmission and queue lengths are monitored in real time. If the length of a data queue continues to increase (such as the fault diagnosis data queue), the token generation rate for the corresponding token bucket is dynamically increased, shortening the refresh cycle. If high-priority data transmission is completed and the queue length decreases, the generation rate is reduced to avoid bandwidth waste.

[0057] This approach dynamically allocates CAN bus bandwidth for chassis control, body control, and fault diagnosis control using a token bucket algorithm, effectively resolving the issue of irrational bandwidth resource allocation. A weighted coefficient and dual-bucket tiering mechanism ensure that high-priority, high-urgency data receives bandwidth first, guaranteeing the real-time transmission of critical data. For example, timely transmission of fault diagnosis information can quickly trigger safety measures, and timely transmission of chassis emergency control signals can prevent accidents. A dynamic adjustment mechanism (adjusting the refresh period and token generation rate based on queue length) allows bandwidth allocation to adapt to the dynamic changes in data volume across each system, improving bandwidth resource utilization and avoiding waste or shortages caused by fixed allocations.

[0058] In related technologies, the real-time data required by the chassis control model and the body control model are transmitted together, which can easily lead to excessive instantaneous data transmission traffic and load overload. Overload can cause data transmission delays and losses, and even affect the normal operation of the chassis control model and the body control model, thereby affecting the vehicle's control accuracy and safety.

[0059] To solve the above problem, in some embodiments, the domain controller further includes: detecting the load of instantaneous traffic through a preset smoothing factor: if it is detected that any load transmitted to the chassis control model and the body control model is greater than the preset load, overload is determined, and the operation frequency of the overloaded data is reduced based on the data priority.

[0060] Exemplarily, a smoothing factor and a preset load are pre-set. The smoothing factor is a parameter used to smooth the instantaneous traffic, and its value range is usually between 0 and 1, for example, it is set to 0.3. The preset load is the threshold for judging whether it is overloaded, which is determined according to the performance and actual needs of the vehicle data transmission system, for example, it is set to 80% of the system's maximum carrying capacity. The instantaneous traffic data transmitted to the chassis control model and the body control model is collected in real time. The formula for smoothing can be expressed as: smoothed traffic = smoothing factor × current instantaneous traffic + (1-smoothing factor) × last smoothed traffic. For example, the current instantaneous traffic is 100Mbps, the last smoothed traffic is 80Mbps, and the smoothing factor is 0.3, then the current smoothed traffic = 0.3×100+(1-0.3)×80=30+56=86Mbps. Through this processing, the sharp fluctuations of instantaneous flow can be reduced, and the actual flow load situation can be reflected more accurately. The smoothed flow is compared with the preset load. If the smoothed flow is greater than the preset load, it is determined that the load transmitted to the chassis control model or the body control model is overloaded; if it is less than or equal to the preset load, it is judged that the load is normal.

[0061] For example, when an overload is detected, the data transmitted to the chassis control model and the body control model that causes the overload is identified and prioritized according to preset data priority rules. Typically, critical chassis control data (such as brake pressure and wheel speed) has the highest priority, followed by important body control data (such as body attitude angle), and non-critical fault diagnosis data has a lower priority. The frequency of overload data is reduced in ascending order of priority. For example, for the lowest-priority non-critical fault diagnosis data, its transmission frequency can be reduced from 10 milliseconds to 50 milliseconds. For slightly higher-priority non-critical body control data, its sampling and transmission frequency can be appropriately reduced. During this frequency reduction process, the frequency of high-priority data is ensured to be unaffected or only slightly adjusted. After reducing the frequency of overload data, the load transmitted to the chassis control model and the body control model continues to be monitored. If the load returns to below the preset value, the current frequency is maintained. If the overload condition persists, the frequency of lower-priority data is further reduced until the load returns to normal.

[0062] This approach effectively detects instantaneous traffic loads and handles overloads by reducing the frequency of overloaded data operations based on data priority, addressing issues such as transmission delays and losses caused by data overload during fault diagnosis. By detecting loads using a preset smoothing factor, the accuracy and reliability of overload detection are improved, avoiding false positives and missed detections. Handling overloads based on data priority ensures the proper transmission of high-priority critical chassis and body control data, safeguarding the stable operation and safety of the vehicle control system.

[0063] In related technologies, during the operation of a vehicle, key target objects such as the sensor group of the data acquisition layer, the actuator group of the instruction execution layer, the chassis control model, and the body control model may fail due to various reasons. If these failures cannot be monitored and handled in a timely manner, the vehicle control system will fail, affecting the normal operation of the vehicle and even causing safety accidents.

[0064] To solve the above problem, in some embodiments, the processing decision layer further includes: A fault diagnosis module is used to monitor the operating status of a target object in the vehicle and determine whether the target object is faulty based on the operating status; if a fault is detected in the target object, a fault alarm message is generated and the backup control scheme corresponding to the faulty object is switched to respond; if no fault is detected in the target object, monitoring is continued. The target object includes at least one of a sensor group in a data acquisition layer, an actuator group in an instruction execution layer, a chassis control model, and a body control model.

[0065] For example, monitoring parameters are determined for different target objects. For sensor groups, these include sensor output signal value, signal stability, and response time. For actuator groups, these include actuator motion accuracy, motion time, operating current, and operating temperature. For chassis and body control models, these include model calculation time, the rationality of output commands, and communication status with other modules. The fault diagnosis module collects monitoring parameter data for the target object in real time via the vehicle's CAN bus and compares the monitored parameters with preset normal ranges. If the parameters are within the normal range, the target object is deemed to be operating normally. If the parameters are outside the normal range, the state is marked as abnormal, and the degree and trend of the abnormality are further analyzed.

[0066] Exemplarily, preset fault judgment rules include single parameter threshold rules, multi-parameter joint rules and trend analysis rules. Single parameter threshold rules: when a certain monitoring parameter exceeds the preset fault threshold, it is judged as a fault; multi-parameter joint rules: when multiple related monitoring parameters simultaneously exceed their respective normal ranges, it is judged as a fault; trend analysis rules: when a monitoring parameter continuously deviates from the normal range for a period of time and the degree of deviation gradually increases, it is judged as a fault. Match the operating status assessment results of the target object with the fault judgment rules. If it meets the fault judgment rules, it is determined that the target object has a fault; if it does not meet the rules, it is judged that no fault has occurred. Optionally, for the target object judged to have a fault, the fault level is divided according to the severity of the fault, such as minor faults, general faults and major faults. Minor faults have little impact on vehicle operation, general faults can affect vehicle performance, and major faults directly threaten driving safety.

[0067] For example, upon determining a target object fault, the fault diagnosis module generates a fault alarm message, including the fault object type, fault occurrence time, fault severity, and fault description. For example, "At a certain time in a certain month in a certain year, a brake pressure sensor in the sensor group of the data acquisition layer experienced a serious fault, resulting in an abnormal output signal." This fault alarm message is transmitted via the vehicle's internal communication network to relevant modules, such as the onboard display, driver information system, and remote monitoring platform, alerting the driver and other personnel. The onboard display displays the fault information in text and an icon, and the driver information system can alert the driver with an audible or visual alarm. Based on the fault object type and fault severity, a pre-defined backup control scheme is invoked. A backup control scheme is a pre-defined alternative control strategy for different target object faults. For example, if a brake pressure sensor fails, the backup control scheme can estimate brake pressure using data from other related sensors (such as wheel speed sensors). If an actuator fails, the backup control scheme can activate a redundant actuator or adjust control instructions to allow another actuator to share its functions. If a control model fails, the backup control scheme can switch to a simplified alternative model or a control strategy based on pre-set rules.

[0068] Through this approach, the fault diagnosis module monitors the operating status of key vehicle targets in real time, identifies faults, and responds to them, effectively resolving the issue of delayed detection and resolution of target target faults. Real-time monitoring and multi-rule fault diagnosis improve the accuracy and timeliness of fault detection, ensuring early detection of faults. Fault alarms provide drivers and other personnel with timely information and the ability to implement countermeasures. Switching to a backup control scheme ensures that the vehicle maintains a certain level of control in the event of a target target fault, preventing loss of control and potential accidents.

[0069] In related technologies, vehicle chassis and body control models are typically designed and trained and calibrated based on specific vehicle parameters, road conditions, and operating environments. However, in actual use, vehicles wear out and age, and the driving environment is complex and ever-changing. These factors can cause the preset control model parameters to gradually deviate from the actual situation, resulting in a decrease in control accuracy.

[0070] To solve the above problems, in some embodiments, the method further includes: The optimization update layer is used to associate and map the received vehicle actual feedback data with the target control instructions to form an optimized data set, and optimize the chassis control model and the body control model based on the optimized data set to update the internal parameters of the chassis control model and the body control model. The vehicle actual feedback data is the actual data generated after the actuator group executes the target control instruction.

[0071] For example, the optimization update layer receives two types of data: the target control instructions generated by the domain controller, such as suspension stiffness adjustment instructions output by the chassis control model and anti-roll bar adjustment instructions output by the body control model; and the actual vehicle feedback data generated after the actuator group executes these target control instructions, such as actual suspension displacement, actual body roll angle, and actual braking deceleration. The target control instructions are mapped to the corresponding actual vehicle feedback data, establishing a one-to-one correspondence based on timestamps or instruction numbers, ensuring that each target control instruction is matched to the actual feedback result after its execution.

[0072] For example, the optimization objectives for the chassis and body control models are clearly defined, typically centered around minimizing the deviation between actual feedback data and the desired target. For example, the optimization objective for the chassis control model might be "the difference between the actual braking deceleration and the target deceleration is less than or equal to 0.5 meters per second squared," and the optimization objective for the body control model might be "the difference between the actual body roll angle and the target roll angle is less than or equal to 2 degrees." Based on the optimization dataset, the relationship between the internal parameters of the current control model (such as the brake pressure coefficient and suspension stiffness coefficient in the chassis control model, and the anti-roll bar force coefficient in the body control model) and the actual feedback deviation is analyzed. Through regression analysis and sensitivity analysis, the key parameters with the greatest impact on control effectiveness are identified. For example, if the deviation between the actual suspension displacement and the target displacement varies primarily with the suspension stiffness coefficient, this parameter is the key optimization target. Optimization algorithms, such as gradient descent and particle swarm optimization, are then used to adjust these key parameters. For example, using gradient descent, the gradient of the deviation function (the difference between the actual feedback and the desired target) with respect to the parameter is calculated. The parameter is then adjusted incrementally along the negative direction of the gradient until the deviation function reaches its minimum. For example, if the actual braking deceleration is less than the target value, the brake pressure coefficient is increased by 0.01 at a time through gradient calculation until the deviation meets the requirements. The optimized parameters are then written into the chassis control model and the body control model to complete the model update. The new target control command and actual feedback data are selected for verification. If the deviation is within the preset range, such as less than or equal to the optimization target threshold, the update is confirmed to be valid. If it does not meet the requirements, the parameter identification and optimization steps are repeated until the model performance meets the requirements.

[0073] Through this approach, the chassis and body control models are dynamically optimized through the optimization update layer, effectively resolving the disconnect between model parameters and actual operating conditions. By constructing an optimized dataset, the model optimization is backed by reliable, real-world data, avoiding blind adjustments based on theoretical assumptions. The parameter optimization process is highly targeted, accurately improving the model's control accuracy in critical scenarios. In practical applications, this solution enables the vehicle control system to self-learn. When the vehicle is new, the model quickly adapts to the vehicle's state using initial data. During use, the model continuously optimizes as the vehicle wears and the environment changes, maintaining a high level of control.

[0074] In some embodiments, see Figure 3 , is a schematic diagram of the principle of the vehicle body chassis collaborative control system provided in an embodiment of the present application, which is described in detail as follows: The chassis-body collaborative control system based on the domain-centralized architecture is based on the domain-centralized architecture. Its specific framework is as follows: The data acquisition layer is composed of a sensor group, including wheel speed sensors, suspension displacement sensors, steering angle sensors, brake pressure sensors, and inertial measurement units. These sensors are installed at corresponding positions on the vehicle and are used to collect chassis and body data (i.e., body status data) such as wheel speed, suspension displacement, steering angle, brake pressure, body posture angle, lateral acceleration, and longitudinal acceleration in real time. The data (body data and chassis data) are then transmitted to the domain controller via the CAN bus.

[0075] The decision-making process includes: The domain controller, the system's core hub, receives data collected by the sensor group and transmits it to the chassis control model and the body control model. It also receives control commands (chassis commands and body commands) generated by the two intelligent models, with chassis commands serving as the primary control command and body commands serving as the secondary control command. The data fusion processing module performs a weighted summation fusion process on these commands before sending the fused target control command to the actuator group. Furthermore, the decision-making layer incorporates a built-in fault diagnosis module to monitor the operating status of each system component in real time and address any faults.

[0076] The chassis control model includes a data preprocessing unit, a feature extraction unit, a strategy matching unit, and an instruction generation unit. It preprocesses and extracts features from the received chassis status data, matches the control strategy, and then generates chassis control instructions in combination with a deep learning algorithm.

[0077] The body control model consists of a data cleaning unit, a pattern recognition unit, a strategy selection unit and an instruction output unit. It cleans and recognizes the body data, selects the control strategy and generates the body control instructions.

[0078] Instruction execution layer The actuator group includes brake actuators, steering actuators, suspension adjustment actuators, and body stability actuators. It receives target control instructions sent by the domain controller, adjusts relevant components of the chassis and body, and realizes chassis-body coordinated control.

[0079] Optimized update layer The chassis control model and body control model use online learning to optimize and update the chassis control strategy, its own control strategy, and algorithm parameters based on the actual driving feedback data of the actuator group in the vehicle, thereby improving system performance through model optimization.

[0080] Through this approach, a domain-centralized architecture replaces the traditional distributed architecture, with the domain controller serving as the core hub to achieve efficient integration and distribution of multi-source data. This architecture breaks the independent model of chassis and body control units, providing a basic framework for chassis-body collaborative control, significantly reducing the complexity of data transmission and the decentralized nature of control logic. Chassis and body control models are constructed, integrating data preprocessing, feature extraction, strategy matching, and command generation, respectively. Using deep learning algorithms, convolutional neural networks, recurrent neural networks, and other technologies, chassis and body data are deeply processed and feature-mined, enabling precise analysis and generating target control commands. Compared to traditional control methods that rely on preset fixed rules, this significantly improves control accuracy and intelligence.

[0081] In some embodiments, a vehicle body chassis coordinated control method is provided, which is used to implement the vehicle body chassis coordinated control system provided in any of the above embodiments. Figure 6 , Figure 6 A flow chart of the vehicle body chassis collaborative control method provided in an embodiment of the present application is detailed as follows: Step S610, collecting chassis data and body data of the vehicle; Step S620: Receive chassis data and body data using a domain controller, forward the chassis data to a chassis control model, and forward the body data to a body control model; perform weighted fusion on a first control instruction from the chassis control model and a second control instruction from the body control model to generate a target control instruction; Step S630: In response to the target control instruction, the vehicle body and chassis are coordinated and controlled.

[0082] In step S610, i.e., step 1, the sensor group includes multiple sensors such as wheel speed and suspension displacement, which are installed at key locations such as wheels and suspensions. They collect chassis (wheel speed, steering angle, etc.) and body data (attitude angle, lateral acceleration, etc.) in real time and transmit them to the domain controller via the CAN bus, thus achieving all-round multi-source data collection. In step S620, i.e., step 2, after receiving the sensor group data, the domain controller transmits it to the chassis control model and the body control model respectively, using a domain centralized architecture to efficiently distribute data and reduce transmission complexity, as detailed below: Step 201: Data frame parsing and protocol conversion. The domain controller processes sensor data in parallel through a multi-channel CAN receiver module, and uses a dynamic frame header recognition algorithm to parse different sensor protocols, such as SAEJ1939 and ISO11898, and converts the original CAN message into a unified internal data structure, supporting a data throughput of up to 1Mbps.

[0083] Step 202: Data priority classification and caching strategy. Based on data timeliness requirements, a three-level caching mechanism is established. For example, first, safety data such as wheel speed and brake pressure enter the SRAM cache with an access delay of less than 100 microseconds; second, posture data is stored in DRAM to support burst reads; third, non-urgent data is written to NAND Flash.

[0084] Step 203: Intelligent routing decision-making. The domain controller has a built-in real-time operating system (RTOS) that implements data diversion through a priority scheduling algorithm. Chassis-related data (wheel speed, steering angle) is transmitted to the chassis control model via a dedicated DMA channel, while body data (acceleration, attitude angle) is distributed to the body control model in a targeted manner, achieving hardware-level isolation of data transmission. See also Figure 4 , is a data partition state transition diagram in the vehicle body chassis collaborative control system provided by an embodiment of the present application; details are as follows: Input: Raw data collected by the sensor group (such as body data, chassis data), data priority label (determined by risk assessment or data type, such as chassis data has higher priority than body data); Output: Data stream distributed to the chassis control model / body control model, hardware-level data transmission path.

[0085] The specific training process is as follows: 1) Initialize the priority queue to an empty set, for example, upon receiving an input signal, or taking a trigger signal as input; 2) Determine the initial vertex for graph growth; Among them, the initial vertex is the CAN bus data receiving port of the domain controller; 3) Add the tuple (data ID, priority, target model) to the priority queue and mark it as queued; examples of tuples are (brake pressure data, high priority, chassis control model) and (body roll angle data, medium priority, body control model); 4) When the queue is not empty, perform the following operations; 5) Pop a tuple from the queue, assuming its vertex is {current data}; Pop up in priority order (high priority data is processed first); 6) Determine whether the partition needs to be started; Partition logic starts the corresponding transmission channel according to the data target model (chassis / body); for example, chassis data starts DMA channel 1 to join the chassis partition, and body data starts DMA channel 2 to join the body partition; 7) Add the vertex (current data) to the current partition node set; 8) Detect the associated data in the current partition node set; 9) If the associated data has not been queued before, it is added to the queue. On the contrary, if there is no unprocessed associated data, it means that the associated data has been processed. By checking whether the data has been processed, duplicate enqueueing is avoided. Add a tuple (association data ID, priority, target model) to the priority queue and mark it as enqueued; Recursively process related data to ensure that data of the same type is transmitted continuously according to priority; return to step 4) until the queue is empty.

[0086] Step 204, load balancing and bandwidth reservation, uses a token bucket algorithm to dynamically allocate CAN bus bandwidth, reserving 60% of the bandwidth for chassis control, 30% for body control, and 10% for fault diagnosis. When the load of a certain model is too high, the adaptive frequency reduction strategy is automatically triggered to ensure the cycle stability of the key control loop (±5%).

[0087] Step 205: Data synchronization and timestamp alignment. Based on the IEEE1588 precision time protocol, nanosecond timestamps are added to all sensor data. Time synchronization is performed through a Kalman filter before data distribution. Data with different sampling rates are aligned to a unified time base, and the synchronization error is less than 500ns. In step S620, i.e., step 3, the chassis control model is filtered, denoised, and normalized by the data preprocessing unit. The feature extraction unit uses a convolutional neural network to extract features. The strategy matching unit matches the control strategy. The instruction generation unit combines deep learning to generate chassis control instructions, and uses a deep learning algorithm to achieve accurate analysis. Input: Chassis status data such as wheel speed, suspension displacement, steering angle, and brake pressure transmitted by the domain controller. Output: Chassis control commands for adjusting chassis-related components.

[0088] S301, the data preprocessing unit, receives the raw chassis status data and filters it using an adaptive median filter algorithm to effectively remove impulse noise. It also uses wavelet denoising technology to eliminate high-frequency interference signals. It then uses the minimum-maximum normalization method to uniformly map the data to the [0,1] interval, improving the stability and efficiency of subsequent data processing. S302, the feature extraction unit, builds a multi-layer convolutional neural network (CNN) model based on the preprocessed data. The convolution kernels in the convolution layer automatically extract local features from the data, such as wheel speed trends and suspension displacement fluctuations. The pooling layer reduces the dimensionality of the features to reduce computational complexity. The fully connected layer integrates the extracted features to form a feature vector suitable for subsequent processing, enabling in-depth mining of chassis status data features. S303, the strategy matching unit performs similarity matching on the feature vector output by the feature extraction unit with the strategies in the preset chassis control strategy library, and uses the cosine similarity algorithm to calculate the similarity between the feature vector and each strategy feature template to find the control strategy with the highest similarity.

[0089] S304, the instruction generation unit, based on the control strategy determined by the strategy matching unit, combines deep learning algorithms such as long short-term memory networks, considers the time series characteristics of the data, analyzes the current chassis status and change trends, generates precise chassis control instructions, such as brake pressure adjustment value, steering angle correction value, etc., and sends them to the domain controller.

[0090] In step S620, i.e., step 4, the body control model removes outliers and interpolates missing values ​​through the data cleaning unit, the pattern recognition unit uses a recurrent neural network to identify motion patterns, the strategy selection unit selects a suitable strategy, and the instruction output unit generates a body control instruction. Input: Vehicle data such as vehicle posture angle, lateral acceleration, and longitudinal acceleration transmitted by the domain controller. Output: Vehicle control commands for adjusting vehicle-related components.

[0091] S401, the data cleaning unit, uses statistical methods to identify and remove outliers from the received vehicle body data, such as data points that exceed three times the standard deviation; for missing values, linear interpolation or a time series-based prediction model is used to interpolate to ensure data integrity and accuracy. S402, the pattern recognition unit, uses a recurrent neural network (RNN), specifically a gated recurrent unit (GRU), to analyze the cleaned data. The GRU can effectively process time series data and capture the dynamic changes in the vehicle body motion pattern, such as the motion pattern characteristics under different working conditions such as vehicle turning, acceleration, and braking, to achieve accurate recognition of the vehicle body motion pattern. S403: The strategy selection unit selects an appropriate control strategy from a pre-set vehicle body control strategy library based on the vehicle body motion mode determined by the pattern recognition unit. The strategy library is categorized by motion mode and uses the correspondence between modes and strategies to quickly select the control strategy that best suits the current vehicle state.

[0092] S404, the instruction output unit, based on the selected control strategy, uses a deep learning algorithm to generate specific body control instructions, such as the intervention force of the body stability actuator, the adjustment parameters of the suspension stiffness and damping, etc., and sends them to the domain controller.

[0093] In step S620 (step 5), the domain controller weights and sums the chassis and body control commands according to preset weights and sends the combined commands to the actuator group. The brake, steering, and other actuators in the actuator group adjust the chassis and body components according to the commands, achieving coordinated control. If the fault diagnosis module built into the domain controller detects a fault in the sensor group, intelligent model, or actuator group, it generates a fault alarm message, returns to step 1, and adjusts the system according to the preset strategy or switches to a backup control mode. If no fault is detected, the system continuously monitors the vehicle's operating status and returns to step 1, where the chassis and body control models optimize the control strategy and algorithm parameters through online learning based on driving feedback data, continuously improving system performance.

[0094] In step 204, the chassis control model and the body control model each process different types of vehicle status data, each of which has varying degrees of importance and urgency for vehicle control. For example, during emergency braking, brake pressure sensor data requires priority processing to ensure driving safety; during sharp turns, the real-time performance of vehicle posture angle and lateral acceleration data is extremely critical. To address this, a data priority assessment mechanism is introduced into the token bucket algorithm, analogous to the data processing logic of the intelligent model, to further optimize CAN bus bandwidth allocation. Specific improvements are as follows: Generation rate of traditional token bucket algorithm is fixed. In the vehicle control system, the data priority evaluation mechanism is introduced into the token bucket algorithm, and its expression is: (1) Among them, in formula (1), is the base token generation rate (unit: tokens / second), is the safety factor, 0> ≤0.5; is the weight of the mth risk factor, satisfying 、 is the normalized function of the mth risk factor. For example, the risk includes speed risk and turning risk.

[0095] Furthermore, a dual-bucket hierarchical token mechanism is set up, introducing two token buckets: high priority (HP) and low priority (LP), to ensure that critical data is transmitted first: Number of HP bucket tokens: (2) Number of LP bucket tokens: (3) Among them, in formula (2) and (3), are the capacities of high / low priority buckets respectively, are the token generation rates for high / low priority buckets respectively, is the number of HP bucket tokens corresponding to time t, The number of HP bucket tokens corresponding to time t-1, is the refresh cycle; is the number of LP bucket tokens corresponding to time t, The number of LP bucket tokens corresponding to time t-1, min is the minimum value solving function.

[0096] Furthermore, the refresh cycle of the traditional algorithm is fixed, but after improvement, it is dynamically adjusted according to the load: (4) In formula (4), is the average queue length, is the maximum queue length, is the adjustment factor, 0.3≤β≤3.0; if the adjustment factor is too small, frequent filling will lead to excess tokens and waste of resources; on the contrary, if the adjustment factor is too large, request discards will be caused due to response delays, which may easily lead to degraded service quality; r is the token generation rate.

[0097] Assign different weights to different types of data to ensure the stability of key control loops: (5) In formula (5), F is the token allocation amount, Z is the total number of tokens, is the static weight of the jth type of data, such as body control ; is the urgency of the j-th category data; For the Static weights for class data, such as chassis control ; For the The urgency of the data.

[0098] A smoothing factor α is introduced to prevent instantaneous traffic overload, as follows: (6) In formula (6), α is the smoothing factor, and the value range of α is 0.1~0.3; when burst traffic is detected, Automatically reduce to suppress peaks, is the token generation rate at the previous moment, is the token generation rate at the current moment, The effective token generation rate.

[0099] In vehicle control systems, the above improvements can be combined: (7) For the Class data in The token generation rate at the moment, No. The base generation rate of class data, System overall risk coefficient, 0≤ ≤0.3; For the The urgency of the class data at time t, For the The weight of the class data, is the weight of the j-th category data, is the urgency of the j-th type of data at time t, calculated based on multi-sensor fusion.

[0100] The CAN bus bandwidth allocation is optimized by introducing a data priority evaluation mechanism, including improvement measures such as dynamic token generation rate, dual-bucket layered transmission, adaptive refresh cycle, association of weight allocation with urgency, and burst traffic smoothing mechanism. These measures are integrated and applied to achieve the goals of prioritizing the transmission of critical data, efficiently utilizing bandwidth, and improving system robustness.

[0101] Among them, Figure 3 As shown, this application builds a complete and efficient chassis-body collaborative control system based on a domain-centralized architecture. At the system architecture level, the domain controller is used as the core hub to connect the data acquisition layer, processing decision layer, instruction execution layer, and optimization update layer. The data acquisition layer collects chassis and body data in real time through various sensors such as wheel speed and suspension displacement, and transmits it to the domain controller via the CAN bus; in the processing decision layer, the domain controller distributes the data to the chassis control model and the body control model. These two models use deep learning algorithms, convolutional neural networks, recurrent neural networks and other technologies to deeply process the data and generate control instructions. The domain controller then fuses the instructions and sends them to the actuator group; various actuators in the instruction execution layer receive instructions to achieve chassis-body collaborative control; the optimization update layer optimizes the control strategy and algorithm parameters based on driving feedback through online learning.

[0102] See also Figure 5 , which is a schematic diagram of the token bucket algorithm structure in the vehicle chassis collaborative control system provided by the embodiment of the present application, and the token bucket algorithm architecture diagram that introduces the data priority evaluation mechanism, is detailed as follows: Data processing and distribution are key innovations in this application. After receiving data, the domain controller utilizes a combination of improved algorithms, including dynamic priority allocation, multi-queue weighted round-robin scheduling, and QoS-based bandwidth reservation, to implement intelligent routing decisions. For example, data priority is dynamically adjusted based on risk assessment, and a dual-bucket hierarchical token mechanism is combined to ensure that critical data is transmitted first. Adaptive token refresh cycles and a prediction-based prefetching strategy (i.e., a control strategy that uses historical data to predict a future period) are utilized to improve data transmission efficiency and provide accurate data support for intelligent models.

[0103] This application solves problems such as low data transmission efficiency, decentralized control logic, lack of coordination and self-optimization capabilities, and realizes intelligent and efficient data processing and precise control, as well as integrated control with deep coordination. At the same time, through fault monitoring and online learning, the system ensures high reliability operation, significantly improving the vehicle's handling performance, stability and safety, and conforming to the development trend of high performance and intelligence of modern automobiles.

[0104] Through this approach, the domain controller, acting as a data transmission hub, receives multi-source data collected by the sensor group. Using strategies such as dynamic priority allocation and multi-queue weighted round-robin scheduling, it diverts data based on its importance and urgency, accurately transmitting chassis and body data to the corresponding intelligent models, avoiding data transmission chaos and congestion. Simultaneously, optimizations such as a dual-bucket hierarchical token mechanism and an adaptive token refresh cycle further ensure the priority transmission of critical data and the rational use of bandwidth, reducing transmission latency. Hardware-accelerated priority arbitration and a prediction-based prefetching strategy improve transmission efficiency at both the hardware and data prefetching levels, reducing data latency. The coordinated operation of this series of technical measures not only significantly reduces the complexity of data transmission but also provides timely and accurate data support for the chassis and body control models, ensuring that the intelligent models can quickly respond to changes in vehicle operating status and generate precise control commands, thereby achieving efficient chassis-body coordinated control and comprehensively improving vehicle handling performance and driving safety.

[0105] In this embodiment, the body-chassis collaborative control method essentially sets up multiple modules to execute the body-chassis collaborative control system in any of the above embodiments. The specific functions and technical effects can be referred to the above embodiments and will not be repeated here.

[0106] In some embodiments, a vehicle is provided, which includes the body-chassis collaborative control system provided by any of the above embodiments.

[0107] For specific vehicle definitions, please refer to the definitions of the body-chassis collaborative control system above and will not be repeated here. Each module in the above-mentioned vehicle can be implemented in whole or in part through software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the electronic device in hardware form, or can be stored in the memory of the electronic device in software form, so that the processor can call and execute the corresponding operations of each of the above modules.

[0108] In one embodiment, an electronic device is provided. The electronic device may be a server, and its internal structure may be as shown in FIG. Figure 7 As shown. The electronic device includes a processor, a memory, a network interface and a database connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile and / or volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external client via a network connection. The computer program is executed by the processor to implement the functions or steps on the server side of the above method.

[0109] In one embodiment, an electronic device is provided. The electronic device may be a client, and its internal structure diagram may be as follows: Figure 8 As shown. The electronic device includes a processor, memory, network interface, display screen, and input device connected via a system bus. The processor of the electronic device is used to provide computing and control capabilities. The memory of the electronic device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer program in the non-volatile storage medium. The network interface of the electronic device is used to communicate with an external server via a network connection. The computer program is executed by the processor to implement the functions or steps of the client side of the above method.

[0110] In one embodiment, an electronic device is provided, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the following steps are performed: Step S610, collecting chassis data and body data of the vehicle; Step S620, using the domain controller to receive the chassis data and body data, forwarding the chassis data to the chassis control model and forwarding the body data to the body control model; performing weighted fusion on the first control instruction from the chassis control model and the second control instruction from the body control model to generate a target control instruction; Step S630, in response to the target control instruction, performing coordinated control on the vehicle body and chassis.

[0111] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the following steps are implemented: Step S610, collecting chassis data and body data of the vehicle; Step S620, using the domain controller to receive the chassis data and body data, forwarding the chassis data to the chassis control model and forwarding the body data to the body control model; performing weighted fusion on the first control instruction from the chassis control model and the second control instruction from the body control model to generate a target control instruction; Step S630, in response to the target control instruction, performing coordinated control on the vehicle body and chassis.

[0112] It should be noted that the above functions or steps that can be implemented by the computer-readable storage medium or electronic device can be referred to the relevant descriptions on the server side and the client side in the aforementioned method embodiment. To avoid repetition, they will not be described one by one here.

[0113] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The above-described computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the above-described method embodiments. Any reference to memory, storage, database, or other media used in the various embodiments provided herein may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), direct memory bus RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0114] Those skilled in the art will clearly understand that for the sake of convenience and brevity in description, only the division of the above-mentioned functional units and modules is used as an example. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the above-mentioned device or system can be divided into different functional units or modules to complete all or part of the functions described above.

[0115] The above embodiments are merely illustrative of the principles and effects of this application and are not intended to limit this application. Anyone skilled in the art may modify or alter the above embodiments without departing from the spirit and scope of this application. Therefore, all equivalent modifications or alterations made by one of ordinary skill in the art without departing from the spirit and technical concepts disclosed in this application shall be covered by the claims of this application.

Claims

1. A vehicle body chassis coordinated control system, characterized in that: include: Data collection layer, used to collect chassis data and body data of the vehicle; a processing decision layer, comprising a centralized domain controller, the domain controller being configured to forward the chassis data to a chassis control model and the body data to a body control model; and further configured to perform weighted fusion of a first control instruction from the chassis control model and a second control instruction from the body control model to generate a target control instruction; An instruction execution layer, configured to perform coordinated control of the vehicle body and chassis in response to the target control instruction; Among them, the domain controller caches the chassis data and the body data in a hierarchical manner according to a hierarchical caching strategy; the domain controller determines the sending order of each data in the data queue to be transmitted according to the data priority, and the chassis data and the body data use different data queues and transmit data respectively through different direct memory access channels; the hierarchical caching strategy uses multiple memories for caching according to the data priority of the received data, each memory has different read and write speeds and access delays, and the data priority is determined by the data type or data risk assessment, and the priority of the chassis data is higher than that of the body data.

2. The vehicle body chassis coordinated control system according to claim 1, characterized in that: The processing decision layer also includes: The chassis control model is configured to preprocess the chassis data, input the preprocessed chassis data into a preset multi-layer convolutional neural network model for feature extraction, and output a first feature vector; perform similarity matching between the first feature vector and a preset chassis control strategy to determine a first control strategy; employ a deep learning algorithm to analyze the state and change trend of the chassis based on time series characteristics, and generate a first control instruction for adjusting the chassis in combination with the first control strategy, wherein the chassis data includes at least one of wheel speed, suspension displacement, steering angle, and brake pressure; The body control model is used to preprocess the body data, input the preprocessed body data into a preset recurrent neural model for pattern recognition, and output a body motion pattern; perform similarity matching between the body motion pattern and a preset body control strategy to determine a second control strategy; generate a second control instruction for adjusting the body according to the second control strategy and a preset deep learning algorithm; the body data includes at least one of body posture angle, lateral acceleration, and longitudinal acceleration.

3. The vehicle body chassis coordinated control system according to claim 1, characterized in that: The domain controller performs weighted calculation on the first control instruction of the chassis and the second control instruction of the body according to a preset chassis weight and a preset body weight, and generates a target control instruction to be sent to an actuator group, wherein the actuator group includes at least one of a braking actuator, a steering actuator, a suspension adjustment actuator and a body stability actuator.

4. The vehicle body chassis coordinated control system according to claim 1, characterized in that: The domain controller dynamically allocates the CAN bus bandwidth required for data transmission for chassis control, body control, and fault diagnosis control according to a token bucket algorithm; wherein the token bucket algorithm controls traffic by a token generation rate and a token number, and the token bucket algorithm satisfies at least one of the following conditions: The token generation rate increases with the weight coefficient of the data type, and each data type is configured with a different weight coefficient according to the data priority; The token bucket algorithm includes a dual-bucket hierarchical token mechanism of high-priority bucket tokens and low-priority bucket tokens, wherein the high-priority bucket tokens are configured to have a token capacity for high-priority data, and the low-priority bucket tokens are configured to have a token capacity for low-priority data, and the token generation rate corresponding to the high-priority bucket tokens is greater than the token generation rate of the low-priority bucket tokens; The token bucket algorithm dynamically adjusts the refresh period according to the queue length of the data queue; The weight coefficient of the data is positively correlated with the urgency, and the number of tokens is allocated to the data based on the weight coefficient and the urgency.

5. The vehicle body chassis coordinated control system according to claim 4, characterized in that: The domain controller further includes: detecting the load of instantaneous traffic through a preset smoothing factor; if it is detected that any load transmitted to the chassis control model and the body control model is greater than the preset load, determining overload, and reducing the operation frequency of overload data based on the data priority.

6. The vehicle body chassis coordinated control system according to claim 1, characterized in that: The processing decision layer also includes: A fault diagnosis module is used to monitor the operating status of a target object in the vehicle and determine whether the target object is faulty based on the operating status; if it is detected that the target object is faulty, a fault alarm message is generated and a switch is made to a backup control scheme corresponding to the faulty object to respond; if it is detected that the target object is not faulty, monitoring is continued, and the target object includes at least one of the sensor group in the data acquisition layer, the actuator group in the instruction execution layer, the chassis control model, and the body control model.

7. The vehicle body chassis coordinated control system according to any one of claims 1 to 6, characterized in that: Also includes: The optimization update layer is used to associate and map the received vehicle actual feedback data with the target control instruction to form an optimized data set, and optimize the chassis control model and the body control model based on the optimized data set to update the internal parameters of the chassis control model and the body control model. The vehicle actual feedback data is the actual data generated after the actuator group executes the target control instruction.

8. A vehicle body chassis coordinated control method, characterized in that: include: Collect vehicle chassis data and body data; A domain controller is used to receive the chassis data and the body data, forward the chassis data to a chassis control model, and forward the body data to a body control model; a first control instruction from the chassis control model is weightedly fused with a second control instruction from the body control model to generate a target control instruction; wherein, the domain controller performs hierarchical caching on the chassis data and the body data according to a hierarchical caching strategy; the domain controller determines the sending order of each data in the data queue to be transmitted according to the data priority, the chassis data and the body data use different data queues, and transmit the data respectively through different direct memory access channels; the hierarchical caching strategy uses multiple memories for caching according to the data priority of the received data, each memory has different read and write speeds and access delays, the data priority is determined by the data type or data risk assessment, and the priority of the chassis data is higher than that of the body data; In response to the target control instruction, the body and chassis of the vehicle are coordinated and controlled.

9. A vehicle, characterized in that: The vehicle adopts the system according to any one of claims 1 to 7.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the method according to claim 8 is implemented.

Images