New energy vehicle operation optimization method and system based on intelligent network connection

Abstract

The invention relates to the technical field of vehicle control, and provides a new energy vehicle operation optimization method and system based on intelligent networking, and the method comprises the steps: collecting a first operation parameter of a vehicle at a current moment and a road parameter at a next moment through a vehicle-mounted sensor network; wherein the first operation parameter comprises a tire movement value corresponding to vehicle suspension deformation variable data based on a current pavement coefficient, and the road parameter is road feature data of a road to be driven in; according to the first operation parameters and the road feature data, a load mean value model and an acceleration safety threshold value of the current vehicle in the road to be driven in are constructed; according to the load mean value model, a damping adjusting instruction of the vehicle suspension is generated; and generating a driving torque vector distribution scheme according to the acceleration safety threshold.

Description

Technical Field

[0001] The present invention relates to the field of vehicle control technology, and in particular to a new energy vehicle operation optimization method based on intelligent networking and a system thereof. Background Art

[0002] With the rapid popularization of new energy vehicles and the in-depth application of intelligent network technology, vehicle operation optimization technology has gradually become a core research direction for improving energy efficiency, driving safety and dynamic adaptability. Traditional new energy vehicle control strategies are mostly based on the real-time state feedback of the vehicle itself (such as vehicle speed, motor torque, battery SOC, etc.), and local optimization is achieved through preset suspension damping adjustment or torque distribution algorithm. However, such methods have significant limitations when dealing with complex road conditions: on the one hand, the dynamic response of the suspension system is delayed, and it is difficult to adjust the damping parameters in advance according to the characteristics of the road ahead (such as slope, curvature, and sudden changes in road adhesion coefficient), resulting in an increase in the tire contact footprint offset, affecting vehicle stability and energy recovery efficiency; on the other hand, the drive torque distribution scheme mostly depends on the current vehicle posture parameters (such as yaw angle, center of mass offset), lacks the ability to predict the load distribution and acceleration safety boundary of the road ahead, and is prone to cause motor overload or tire slip loss of control under continuous slope changes or sharp turns. In addition, the existing technology lacks coordinated optimization of suspension control and drive systems, especially in the real-time matching of multi-source sensor data fusion (such as LiDAR point cloud, MEMS array signal) and cloud-based high-precision maps. A global optimization framework based on road preview has not yet been formed, resulting in the robustness and real-time performance of control commands being difficult to meet the requirements of complex scenarios.

[0003] Although current research attempts to introduce intelligent network technology to improve the above problems, such as obtaining road information ahead through V2X communication, key technical bottlenecks still exist: First, the processing of road excitation response signals mostly adopts a single frequency domain analysis method, which fails to effectively separate the nonlinear coupling relationship between the tire resonance frequency band and the suspension dynamic deformation characteristic quantity, resulting in a large prediction error of the tire movement value; second, the construction of the load mean model is usually based on the static sprung mass assumption, ignoring the dynamic influence of the road curvature and slope gradient on the unsprung mass, especially when the vehicle enters a compound curvature road, the model prediction accuracy drops sharply; third, the torque distribution scheme mostly adopts a fixed threshold constraint, and fails to combine the motor temperature, battery SOC state and tire slip rate gradient to construct a dynamic feasible domain, resulting in the actuator being prone to saturation risk. In addition, the existing methods lack a multi-objective game mechanism in the coordinated optimization of damping adjustment and torque distribution. When the longitudinal acceleration exceeds the safety threshold, it is difficult to achieve nonlinear optimization of the four-wheel load ratio by adjusting the front / rear axle torque distribution weight coefficient in real time, which in turn affects the comprehensive energy efficiency and handling stability of the vehicle. Summary of the invention

[0004] This application proposes a method and system for optimizing the operation of new energy vehicles based on intelligent networking to address the problems of suspension dynamic adjustment lag, insufficient robustness of drive torque distribution scheme, and lack of multi-source data fusion and road preview capabilities in the prior art of new energy vehicles under complex road conditions. Through the nonlinear mapping of three-dimensional road excitation response signal decomposition and dynamic compensation factors, combined with rolling time domain calibration of historical trajectory data of adjacent vehicles, the prediction accuracy of tire movement values ​​is improved, and the reliability of distribution under complex working conditions is improved.

[0005] In the first aspect, the present application proposes a method for optimizing the operation of new energy vehicles based on intelligent networking, including:

[0006] The first operating parameter of the vehicle at the current moment and the road parameter at the next moment are collected through the vehicle-mounted sensor network; wherein the first operating parameter includes the tire movement value corresponding to the vehicle suspension deformation variable data under the current road surface coefficient, and the road parameter is the road characteristic data of the road to be entered;

[0007] Constructing a load mean value model and an acceleration safety threshold value of the current vehicle on the road to be entered according to the first operating parameter and the road characteristic data;

[0008] Generate a damping adjustment instruction for the vehicle suspension according to the load mean model;

[0009] Generate a drive torque vector distribution plan based on the acceleration safety threshold.

[0010] In this application, by collecting the current vehicle operating parameters and the road parameters at the next moment (such as slope and curvature) in real time, the road preview capability is combined with the vehicle dynamic response, and the load model and safety threshold are constructed in advance to solve the lag problem of traditional control strategies and make the suspension damping adjustment and torque distribution forward-looking. Through the dual constraints of the load mean model and the acceleration safety threshold, the suspension system and the drive system are optimized in coordination, and comfort, stability and energy efficiency are balanced under complex road conditions, avoiding performance conflicts caused by single system optimization.

[0011] In combination with the first aspect, the collecting of the first operating parameter of the vehicle at the current moment and the road parameter at the next moment includes:

[0012] Obtaining a three-dimensional road excitation response signal collected by a wheel hub embedded MEMS array in a vehicle-mounted sensor under a first operating parameter of the vehicle at the current moment, performing three-dimensional road excitation response signal decomposition, and determining a suspension dynamic deformation characteristic quantity;

[0013] Through real-time matching of the on-board LiDAR point cloud data in the on-board sensor with the high-precision map in the cloud, the equivalent curvature radius and slope change gradient of the road within the distance to be optimized ahead are calculated to determine the road parameters at the next moment.

[0014] In this application, the three-dimensional response signal decomposition of the wheel hub MEMS array is used to accurately extract the dynamic deformation characteristics of the suspension, improve the real-time and resolution of road excitation recognition, avoid the noise interference caused by the overlap of the traditional single sensor signal frequency band, and calculate the equivalent curvature radius and slope change gradient through real-time matching of LiDAR point cloud and cloud-based high-precision map, solve the deviation problem between static map data and dynamic road characteristics in the traditional preview algorithm, and enhance the robustness of model prediction.

[0015] In combination with the first aspect, the tire movement value collection process includes:

[0016] Perform frequency domain analysis on the suspension dynamic deformation to extract the target energy distribution in the preset characteristic frequency band associated with the tire resonance frequency;

[0017] Generate a dynamic compensation factor based on the discrete characteristic parameter of the road ahead; wherein the dynamic compensation factor includes a weighted combination of the road curvature and the slope change rate;

[0018] Nonlinearly mapping the target energy distribution with the dynamic compensation factor to generate a correction value for the tire contact patch offset;

[0019] The historical trajectory data of adjacent vehicles is obtained through on-board V2X communication, and the correction value is calibrated in the rolling time domain to generate a tire movement value that integrates multi-source information.

[0020] In this application, the energy distribution of the tire resonance frequency band is extracted through frequency domain analysis, and the dynamic compensation factor (such as curvature and slope weighted combination) generated by the road discreteness characteristic parameters is combined to solve the problem of nonlinear error accumulation caused by road mutations in traditional tire movement value calculation. The rolling time domain calibration of the historical trajectory data of adjacent vehicles is introduced to reduce the noise interference of single vehicle sensors and improve the anti-disturbance ability of tire movement value prediction.

[0021] In combination with the first aspect, the load mean model is generated based on a hybrid observer model including energy exchange between sprung mass and unsprung mass; wherein the hybrid observer model performs online correction under road characteristic data and performs parameter adaptation according to the online correction.

[0022] In this application, the energy exchange model between sprung mass and unsprung mass is used to capture the energy transfer characteristics between the suspension and the tire in real time, solve the error problem of ignoring the dynamic influence of unsprung mass in the traditional linear model, and perform online correction on the model based on road characteristic data to improve the prediction stability of the load mean model under continuously variable curvature and variable slope road conditions, and avoid control command oscillation caused by model mismatch.

[0023] In combination with the first aspect, the construction of the load mean model and the acceleration safety threshold of the current vehicle on the road to be entered further includes:

[0024] When the predicted value of the load mean exceeds the standard load range, the multi-objective optimization algorithm is started to calculate the optimal stiffness sequence of the damper;

[0025] A fuzzy PID controller is used to generate a nonlinear adjustment curve of the damping valve opening; wherein the nonlinear adjustment curve includes a dynamic feedforward compensation term based on the slope change rate.

[0026] In this application, when the load exceeds the limit, the optimal stiffness sequence of the damper is calculated by a multi-objective optimization algorithm, combined with the dynamic feedforward compensation of the fuzzy PID controller to solve the overshoot and response lag problems caused by sudden slope changes in traditional PID regulation. Based on the slope change rate, a nonlinear regulation curve is generated to achieve feedforward-feedback composite control of the damping valve opening, thereby enhancing the transient response capability of the suspension system under sudden road conditions.

[0027] In combination with the first aspect, the constructing of the load mean model of the current vehicle on the road to be entered further includes:

[0028] Pre-configure the motor temperature constraints of current new energy vehicles and associate them with the battery SOC state to form a feasible domain for torque distribution;

[0029] Based on the model predictive control framework, the optimal torque distribution solution that satisfies the tire slip ratio constraint is solved in the torque distribution feasible domain.

[0030] In this application, the motor temperature is associated with the battery SOC state to construct a feasible domain for torque distribution, avoiding the risk of actuator failure caused by motor overheating or battery overdischarge in traditional torque distribution. The optimal torque distribution solution is solved based on the model predictive control framework to ensure that the tire slip rate is always within the safety threshold and solve the problem of drive wheel slippage under rapid acceleration or low-adhesion roads.

[0031] In combination with the first aspect, generating a damping adjustment instruction for a vehicle suspension according to a load mean value model includes:

[0032] Obtain the vertical acceleration of the sprung mass of the current vehicle and construct a dynamic load distribution coefficient matrix;

[0033] According to the dynamic load distribution coefficient matrix, under the damping reference prediction, an initial damping coefficient sequence is generated, and a tire slip rate change gradient is obtained;

[0034] When it is detected that the vehicle is on a preset complex road, the motor torque distribution characteristics are determined based on the tire slip rate change gradient, and the response bandwidth of the damping correction command is dynamically adjusted.

[0035] In this application, the response bandwidth of the damping correction command is adjusted according to the gradient of the tire slip rate change, so as to reduce high-frequency vibration interference on complex roads (such as icy and snowy roads) and improve the stability of suspension adjustment. The initial damping sequence is generated through the dynamic load coefficient matrix to avoid the energy imbalance problem between the sprung mass and the unsprung mass caused by the traditional fixed damping coefficient.

[0036] In combination with the first aspect, generating a driving torque vector distribution scheme according to the acceleration safety threshold includes:

[0037] Collect the center of mass projection offset and yaw rate deviation of the current vehicle in real time to generate dynamic feasible domain boundary conditions of the acceleration safety threshold;

[0038] According to the sliding mode variable structure control algorithm, solving the reference torque distribution matrix that satisfies the Lyapunov stability condition in the dynamic feasible domain;

[0039] The actuator saturation risk is identified through the brake system pressure feedback data, and a torque redistribution instruction with anti-saturation characteristics is generated to generate a drive torque vector distribution plan.

[0040] In this application, the feasible domain boundary of the acceleration safety threshold is generated based on the center of mass offset and the yaw angular velocity, solving the risk of actuator saturation or instability caused by the fixed threshold in traditional sliding mode control. The saturation risk is identified through brake pressure feedback and a torque redistribution command is generated to ensure the feasibility of driving torque distribution in emergency obstacle avoidance or extreme working conditions.

[0041] In combination with the first aspect, generating a driving torque vector distribution scheme according to the acceleration safety threshold further includes:

[0042] When the actual value of the longitudinal acceleration exceeds the safety threshold, the current vehicle torque distribution coefficient based on the multi-objective game algorithm is activated, and the front / rear axle torque distribution weight coefficient is dynamically adjusted. The four-wheel load ratio output by the load mean prediction model is nonlinearly negatively correlated.

[0043] In this application, when the longitudinal acceleration exceeds the limit, the front / rear axle torque weight coefficient is adjusted through a multi-objective game algorithm to solve the problem of unreasonable load distribution between axles caused by traditional equal torque distribution. The four-wheel load ratio is constrained by nonlinear negative correlation to improve the vehicle's pitch stability and grip during sudden acceleration or emergency braking.

[0044] In the second aspect, the present application proposes a new energy vehicle operation optimization system based on intelligent networking, the system comprising:

[0045] Road data collection module: collects the first operating parameter of the vehicle at the current moment and the road parameter at the next moment through the vehicle-mounted sensor network; wherein the first operating parameter includes the tire movement value corresponding to the vehicle suspension deformation variable data under the current road surface coefficient, and the road parameter is the road characteristic data of the road to be entered;

[0046] A load acceleration generation model: used to construct a load mean value model and an acceleration safety threshold value of the current vehicle on the road to be entered according to the first operating parameter and the road characteristic data;

[0047] Vehicle suspension adjustment module: used to generate damping adjustment instructions for vehicle suspension according to the load mean model;

[0048] Torque vector distribution module: used to generate a drive torque vector distribution plan based on the acceleration safety threshold.

[0049] In this application, independent modules are used to realize hierarchical processing of data collection, model building, suspension adjustment and torque distribution, reduce system coupling, improve real-time performance and scalability. Each module is linked with constraints based on a unified data source to form a closed-loop optimization chain from road preview to execution control, solving the problem of low collaborative efficiency caused by information islands in traditional distributed control.

[0050] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description and the accompanying drawings.

[0051] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0052] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0053] Figure 1 A method flow chart of a method for optimizing operation of a new energy vehicle based on intelligent networking in an embodiment of the present invention;

[0054] Figure 2 Schematic diagram of a road information collection process of multimodal sensing and road prediction in an embodiment of the present invention;

[0055] Figure 3 FIG. 1 is a diagram of the implementation process of tire movement value calibration of multi-source data fusion in an embodiment of the present invention;

[0056] Figure 4A schematic diagram of the process of constructing a load mean value model for dynamic energy exchange in an embodiment of the present invention;

[0057] Figure 5 A schematic diagram of a multi-objective optimization process in an embodiment of the present invention;

[0058] Figure 6 It is a process flow chart of torque distribution of multiple constraint coupling in an embodiment of the present invention;

[0059] Figure 7 It is a schematic diagram of the damping control process of the dynamic load combined with the slip rate coupling in an embodiment of the present invention;

[0060] Figure 8 A schematic diagram of a process of a system for dynamic stability and anti-saturation control in an embodiment of the present invention;

[0061] Fig. 9 Schematic diagram of the collaborative control process of nonlinear constraints in an embodiment of the present invention;

[0062] Fig.10 This is a system composition diagram of a new energy vehicle operation optimization system based on intelligent networking in an embodiment of the present invention. DETAILED DESCRIPTION

[0063] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0064] In the traditional new energy control process, the optimization of the operation process mainly focuses on the core areas of power battery management, motor efficiency improvement and regenerative braking. In this process, the coordinated optimization of the vehicle suspension system and the drive torque distribution is relatively weak. Therefore, the traditional technical solution usually adopts the "perception-feedback" control architecture, which realizes local adjustment through preset thresholds or linearization models based on the real-time data of the vehicle's own sensors, but it will face significant performance bottlenecks under complex road conditions.

[0065] Because traditional suspension control is mainly based on sprung mass vertical acceleration sensors or road type recognition algorithms, and then improves vehicle smoothness by adjusting the damping valve opening or air spring stiffness.

[0066] For example, the Skyhook Control or Groundhook Control algorithm is used to adjust the damping force in real time according to the vehicle body acceleration feedback; in some driving scenarios, the road recognition module's camera or accelerometer frequency domain analysis is used to determine the discrete types of roads such as "smooth" and "bumpy", and then switch to the preset damping mode.

[0067] Traditional methods rely on the vehicle state feedback at the current moment, and cannot perceive the sudden change in slope and curvature on the road ahead in advance, which will cause the suspension adjustment command to lag behind the actual road conditions. For example, if the vehicle is about to enter a continuous speed bump, the traditional damping parameters need to be adjusted only after the tire has been impacted, resulting in increased vibration of the sprung mass and reduced energy recovery efficiency. The traditional load mean model is based on the static assumption of the sprung mass, thereby ignoring the dynamic energy exchange effect of the unsprung mass such as tires and wheels. Through the coupling of sprung and unsprung masses in compound curvature roads such as continuous S-bends, the coupled vibration causes the model prediction deviation to expand, and then causes the damping parameters to be over-adjusted or under-adjusted, which is manifested as the body pitch and roll amplitude exceeding the safety threshold. The calculation of the suspension dynamic deformation and tire movement value in traditional technology relies on a single frequency domain analysis, such as FFT transformation, which does not separate the tire resonance frequency band and the suspension low-frequency vibration signal, resulting in high-frequency noise interference in the correction value of the tire contact footprint offset. However, on low-grip surfaces, such as wet or icy roads, such errors can significantly increase the risk of tire slippage.

[0068] The driving torque distribution is usually based on the vehicle's yaw rate, wheel speed difference and the driver's required torque, and the front and rear axle torque distribution is achieved using a rule base or PID controller. For example, the electronic stability program (ESP) monitors signs of vehicle instability (such as excessive yaw angle deviation) to trigger braking intervention or limit the driving torque; some schemes adopt an equal distribution strategy to distribute the total required torque to each drive motor in a fixed proportion. The traditional method does not integrate the parameters of the road ahead (such as slope gradient, curvature radius), resulting in the inability of the torque distribution scheme to predict the load transfer trend. For example, when the vehicle is about to enter a steep slope, the front axle load will increase significantly, but the traditional strategy still divides the torque equally according to the flat road condition, resulting in a surge in the front wheel slip rate and a decrease in motor efficiency. The torque distribution feasible domain is not associated with dynamic parameters such as motor temperature and battery SOC state. In the scenario of continuous climbing or high-load acceleration, the motor overheating protection mechanism may be forcibly triggered, resulting in a sudden drop in torque output, causing a fault in driving experience or even safety hazards.

[0069] Traditional road preview relies on the local environment perception of on-board sensors (such as cameras and millimeter-wave radars), or obtains information such as road curvature and slope through GPS positioning and matching offline high-precision maps. Data fusion often uses Kalman filtering or weighted average algorithms to integrate multi-sensor signals into a single road feature parameter. The update cycle of offline high-precision maps is long (usually quarterly), and it cannot reflect dynamic changes such as temporary construction and road damage, resulting in significant deviations between preview road parameters (such as curvature radius) and actual road conditions. For example, in a curve that is diverted due to construction, preview information based on static maps will mislead the suspension and torque control strategies, increasing the risk of vehicle skidding. Traditional data fusion algorithms are difficult to effectively analyze heterogeneous data from multimodal sensors such as LiDAR point clouds and MEMS arrays. For example, the road excitation signal collected by the wheel hub MEMS sensor is easily contaminated by the electromagnetic noise of the motor, and the signal decomposition without frequency-space domain joint filtering will introduce errors in the calculation of suspension deformation.

[0070] Embodiment 1:

[0071] This application proposes a new energy vehicle operation optimization method based on intelligent networking, see Figure 1 This application uses the suspension displacement sensor, wheel speed sensor, and inertial measurement unit controlled by the vehicle-mounted sensor network to collect operating parameters such as tire movement value and body posture angle in real time. Then, the high-precision map and V2X communication are combined to obtain the characteristic data of the road to be entered, such as road roughness, slope, and curve curvature, to build a cross-temporal and spatial data fusion model through multi-source data. The spatiotemporal data fusion model is not limited to the data of a single moment, and will dynamically associate the road characteristics of the next moment with the current vehicle status through a preview mechanism.

[0072] For example, if the vehicle is about to enter a series of curves, the vehicle system will predict the distribution change trend of the tire ground force based on the curvature of the road ahead and the current suspension compression.

[0073] This application uses the finite element analysis method to convert the tire movement value into the dynamic distribution of the load on each axle of the vehicle. The sensor noise interference is eliminated through the Kalman filter algorithm, and then a load mean field centered on the center of mass of the vehicle is generated. The field reflects the vertical force equilibrium state of each wheel under different road conditions and provides a quantitative basis for suspension control. For example, if a sudden increase in the right front wheel load is detected, which indicates a unilateral impact, the model built in this application automatically marks the area as a high-risk area and triggers the pre-adjustment mechanism.

[0074] This application calculates the dynamic safety boundary of the vehicle's longitudinal / lateral acceleration based on the road adhesion coefficient and load distribution state. In actual implementation, this application adopts the model predictive control (MPC) framework to optimize the vector distribution scheme of the driving torque by ensuring that the acceleration does not cross the boundary. For example, in the scenario of starting on a slippery slope, this application will dynamically limit the motor output torque according to the real-time μ value of the vehicle and the system, and intelligently distribute the torque ratio of the front and rear axles to suppress the risk of slipping.

[0075] In actual testing, this application introduces a vehicle that is about to enter a damaged asphalt road from a smooth highway:

[0076] The vehicle's V2X system obtains the road roughness data ahead 500 meters in advance (the advance distance can be set by yourself), and then detects the current tire ground pressure distribution through the suspension sensor.

[0077] The load mean model of the present application predicts that the damaged road surface will cause an increase in the load fluctuation of the left rear wheel, and the acceleration safety threshold module of the present application calculates the maximum allowable vertical acceleration.

[0078] The active suspension system of this application adjusts the damping of the left rear shock absorber to a medium-high gear 200 meters in advance (the number of meters can also be modified in practice) to prevent the expected impact; then the left rear wheel torque output weight is reduced through the driving torque distribution module, and the body stability is maintained through the right front wheel torque compensation. The vehicle is controlled to pass the damaged road section in a smooth posture, and the driver and passengers do not feel obvious bumps.

[0079] Embodiment 2:

[0080] This application proposes a road acquisition method through multimodal sensing and road prediction, see Figure 2 In this application, in the wheel hub embedded MEMS sensor array, 32 micro-electromechanical units are used to realize spherical distributed layout, and the three-dimensional vibration acceleration signals of the tire contact area and the three-axis three-axis vibration acceleration signals are collected in real time. Then, the excitation components of different frequency bands are separated by wavelet packet decomposition algorithm:

[0081] For example:

[0082] Low-frequency component (0-20 Hz), which is used to reflect the quasi-static deformation characteristics of the suspension system;

[0083] The mid-frequency component (20-100 Hz) is used to characterize the contact impact energy between the tire and the road surface;

[0084] High frequency component (>100 Hz), which is used to indicate the friction characteristics of the tread pattern and micro texture;

[0085] The present application combines the vehicle dynamics model to map the multi-band signal into the dynamic deformation characteristic quantity of the suspension, including: vertical compression rate, lateral deformation gradient, and torsion beam stress distribution.

[0086] For example, if the vehicle passes through continuous speed bumps, the MEMS array configured in this application captures the periodic Z-axis impact signal, and the vehicle system will identify the resonant frequency offset of the suspension dual-mass model through spectrum analysis, thereby triggering the pre-adjustment mechanism.

[0087] The solution in this application enables the vehicle-mounted LiDAR to obtain 3D point cloud data within a range of 150 meters in front in a rotating scanning mode, and then perform real-time registration with the cloud-based high-precision map through the iterative closest point algorithm (ICP):

[0088] By extracting key geometric features such as road edge curvature mutation points and slope turning lines;

[0089] Then, the lane curvature and elevation data are probabilistically matched based on the Gaussian mixture model (GMM);

[0090] In actual implementation, within the distance to be optimized, usually 50-200 meters, the sliding average of the effective curvature radius and the slope change gradient vector will be calculated along the expected trajectory of the vehicle. For example, when there are continuous curves in mountainous areas, the vehicle computer identifies the combined curvature characteristics of the "S"-shaped curve ahead in advance, and then dynamically constructs a continuous change function of the curvature radius.

[0091] This application introduces the scenario where a vehicle encounters a road construction area while driving on a rainy night. If there are irregular potholes and temporary curves in the right lane:

[0092] This application will detect a sudden increase in high-frequency vibration energy through the right front wheel hub MEMS array, reflecting the impact of gravel, and a wide-band random vibration in the mid-frequency band, indicating continuous irregular dents. The signal decomposition module identifies that the lateral deformation gradient of the suspension exceeds the safety threshold.

[0093] This application uses LiDAR point cloud scanning to find that the lane line disappears 30 meters ahead, and then identifies the temporary re-route curve through high-precision map matching, and then calculates the equivalent curvature radius to drop sharply to the dangerous range, and the slope change gradient vector points to the right.

[0094] In this application, the suspension pre-adjustment model increases the damping hardness of the right front suspension and transfers the torque vector to the left rear wheel in advance through the power distribution module. Then the vehicle passes through the complex and rugged road section in a stable posture, thereby avoiding the control instability caused by unilateral suspension overload.

[0095] Embodiment 3:

[0096] This application proposes a tire movement value calibration system that integrates multi-source data. Figure 3 In actual implementation, the time domain signal of the suspension displacement sensor is decomposed in the frequency domain through short-time Fourier transform, and then the characteristic frequency band associated with the tire resonance frequency is identified. In actual implementation, it is usually 8-15Hz. The energy entropy algorithm is used to quantify the energy proportion of each frequency band and construct the target energy distribution matrix. For example, in actual implementation, when the vehicle passes over a speed bump, the suspension deformation has an energy peak in the 10Hz frequency band. The system determines that the tire vertical resonance is triggered and ground contact footprint compensation is required. In this application, based on the equivalent radius of curvature and the slope change rate of the road ahead, in the process of constructing a three-dimensional discrete parameter space, the curvature discreteness is calculated by sliding the window. The coefficient of variation of the curvature radius can reflect the continuity of the curve; the slope discreteness uses the gradient vector field to analyze the consistency of the slope change direction. Then, the two are weighted and fused through the fuzzy logic algorithm to generate a dynamic compensation factor. For example, in multiple continuous S-shaped curves, the curvature discreteness and the slope discreteness are both at a high level, and the compensation factor will automatically increase to cope with complex lateral force changes.

[0097] This application establishes a nonlinear mapping relationship between target energy distribution and dynamic compensation factor. It uses a radial basis function (RBF) neural network model to map input parameters to a high-dimensional space, and generates an offset correction value for the tire contact patch through the output layer to reflect the coupling effect of road surface excitation and road characteristics; then the tire slip rate, steering angle timing, etc. corresponding to the historical trajectory data of adjacent vehicles are obtained through V2X communication, and the rolling time domain estimation algorithm is used to calibrate the correction value in real time. For example, if there are multiple vehicle trajectories showing that a certain curve has a local low adhesion area, the vehicle system automatically increases the compensation weight of the correction value.

[0098] When this application is actually implemented in h, by introducing a road scenario with multiple late arrivals in mountainous areas, when there are vehicles passing through sharp bends and undulating slopes continuously, the right rear suspension continues to have energy pulses in the 12Hz frequency band, and the vehicle system recognizes it as tire lateral resonance, and the risk of the contact patch deviation to the right increases. Then the discreteness of the curvature of the road ahead reaches the dangerous threshold, the slope change gradient points to the left tilt, and the dynamic compensation factor generates a strong right correction instruction. By receiving the trajectory data of the vehicle in front showing that there is a tire slip anomaly at the same position, the vehicle system increases the weight of the correction value by 20%. The torque vector distribution module enhances the left front wheel drive torque, the active suspension adjusts the stiffness of the right rear shock absorber, and the vehicle passes through the complex dangerous section with a stable trajectory.

[0099] Embodiment 4:

[0100] This application proposes a method of dynamic energy exchange and constructing a load mean model. Figure 4First, this application will establish an extended state space equation for a dual-mass spring-damper system, and quantify the kinetic and potential energy exchange process between the vehicle body and the unsprung mass, such as wheels and suspension, corresponding to the sprung mass into an energy flow matrix. Then the Lyapunov function is used to analyze the stability boundary of the energy flow and identify key energy transfer paths, such as vertical vibration energy transfer and lateral kinetic energy dissipation. This application adjusts the transfer function parameters of the observer model in real time based on the equivalent stiffness coefficient and damping characteristics of the road ahead. The recursive least squares method (RLS) is used to identify the road excitation signal online;

[0101] For example, if a vehicle enters a gravel road from an asphalt road, the model of this application automatically enhances the high-frequency vibration suppression weight. The benchmark parameter sets for typical road conditions, such as flat roads, speed bumps, and gravel roads, are pre-stored. This application uses a fuzzy logic algorithm to achieve smooth switching of parameter sets. Within the 5-step prediction time domain, based on vehicle dynamics constraints, such as roll angle limits and tire grip boundaries, the observer gain matrix is ​​optimized to ensure real-time consistency between the model output and the measured data.

[0102] This application introduces a combination of continuous undulating roads and curves, where the vehicle experiences the combined effects of vertical vibration and lateral centrifugal force. The unsprung mass sensor detects the sudden increase in vertical kinetic energy of the right front wheel and the impact of the road potholes, and the sprung mass sensor simultaneously captures the change in the body roll angle rate. The hybrid observer identifies abnormal energy flow paths, for example: vertical kinetic energy is transmitted to the side. This application enhances the roll stiffness parameter weight based on the curvature data of the curve ahead. The load mean model of this application redistributes the vertical loads on the four wheels, the suspension system enhances the damping of the outer shock absorber, and the drive torque module increases the torque output of the inner wheel. Finally, the vehicle passes through the composite section with the minimum roll angle, and the body posture stability is significantly improved.

[0103] Embodiment 5:

[0104] This application proposes a multi-objective optimization method, see Figure 5When the predicted value of the mean load exceeds the standard range, a multi-objective optimization process based on NSGA-II (non-dominated sorting genetic algorithm) is triggered. The optimization objectives of this application include adjusting the minimum vertical vibration acceleration for driving comfort, thereby reducing the minimum vector loss of damper power consumption, and maximizing the tire contact force balance for handling stability. The change rate of the stiffness sequence is not higher than the mechanical structure limit, and the difference between adjacent stiffness values ​​will be suppressed due to the hysteresis effect. The optimal stiffness sequence is output through screening of the Pareto frontier solution set. For example, in a continuous speed bump scenario, the algorithm generates a "high-low-high" alternating stiffness sequence by balancing the impact absorption and energy recovery requirements. Under the setting of fuzzy logic parameters, a three-dimensional fuzzy rule base is constructed to realize the calculation of error, error change rate and slope change rate. By dynamically adjusting the PID parameters, if the proportional term weight is enhanced when the slope change rate is large, the response speed will be improved. If the vertical vibration is severe, the differential term will be enhanced to suppress overshoot. Based on the gradient vector of the real-time slope change rate, an advance compensation term is constructed: under uphill conditions, the preload compensation of the damping valve opening will be increased; under downhill conditions, a negative opening will be introduced to suppress the pitch trend. Finally, the convergence of the compensation term is verified through Lyapunov stability analysis to ensure that the control curve is smooth and free of oscillation.

[0105] When this application is actually implemented, by introducing a situation where the vehicle encounters alternating sections of continuous ramps and damaged roads on mountain roads, the predicted value of the mean load on the right rear wheel exceeds the upper limit, and the system identifies it as a suspension overload risk. The NSGA-II algorithm calculates the stiffness sequence [K1, K2, K3], giving priority to ensuring the balance of tire ground contact. Based on the current slope change rate (+8% gradient), the fuzzy rule base enhances the proportional term weight, and the feedforward module generates a preload compensation curve. The damping valve adjusts the opening in stages according to the nonlinear curve, and the vehicle achieves a body pitch angle of ≤2° at the top of the ramp, and the tire slip rate maintains a safe range.

[0106] Embodiment 6:

[0107] This application proposes how to achieve torque distribution with multiple constraints, see Figure 6 ,Firstly, the temperature-torque derating curve is constructed through the association constraint of the motor temperature and SOC;

[0108] This application establishes an exponential torque derating function based on the motor winding temperature sensor data. If the temperature approaches the heat resistance limit of the material, the slope of the function increases sharply, forming a nonlinear constraint boundary.

[0109] The SOC-power mapping model is a model that dynamically divides the allocatable range of driving power according to the real-time SOC value of the battery management system. In specific implementation, an asymmetric allocation strategy is adopted through the low SOC interval to prioritize the basic power demand.

[0110] A three-dimensional feasible domain is constructed, and then the temperature constraints and SOC constraints are projected into the torque distribution space to generate a dynamically shrinking polyhedral feasible domain.

[0111] For example, under high temperature and low SOC conditions, the feasible domain shrinks significantly in the direction of high torque, forcing the system to enter energy-saving mode. In the 5-step prediction time domain, a multi-objective cost function including tire slip rate, motor efficiency, and battery discharge rate is constructed.

[0112] Then the Pareto optimal solution set is solved by the sequential quadratic programming (SQP) algorithm. Based on the slack variable method to deal with the hard constraint of slip rate, temporary slight slip can be allowed in exchange for the global optimal energy consumption.

[0113] For example, in a snow start scenario, the system briefly allows a 5% slip rate to reduce the peak load on the motor.

[0114] When this application is actually implemented, the vehicle will be introduced into a long-distance hill climbing in a high temperature environment. When the battery SOC drops to the critical interval, the motor winding temperature corresponding to this application reaches 90°C, and the SOC drops to 15%. The control feasible domain shrinks to 40% of the original area. Then the temperature cumulative damage factor is introduced into the cost function to optimize the distribution scheme of 70% front wheel torque and 30% rear wheel torque. Based on the risk of suppressing overheating of the rear wheel motor, the front wheel slip rate is monitored. If the instantaneous value exceeds the threshold, the torque feedforward compensation will be triggered. By transferring excess torque to the non-slipping wheels, the vehicle completes the climbing with smooth power output, the motor temperature is stabilized in a safe range, and the battery discharge rate is always lower than the maximum allowable value.

[0115] Embodiment 7:

[0116] This application proposes a damping control method for coupling dynamic load with slip rate, see Figure 7 ,Firstly, the time-frequency characteristics of the vehicle body vertical ,acceleration are collected through the six-dimensional inertial measurement unit, and ,empirical mode decomposition is used to separate different vibration modes, and ,determine the data such as low-frequency vehicle body rigid motion and high-frequency ,suspension resonance.

[0117] The dynamic equilibrium state of the four-wheel load is quantified by constructing a load distribution coefficient matrix with the wheel load fluctuation amplitude as the horizontal axis and the phase lag angle as the vertical axis.

[0118] By introducing sliding window covariance analysis, based on the changing trend of matrix eigenvalues ​​in real time monitoring, if the difference of main diagonal elements exceeds the threshold, it will be judged as a single wheel overload risk and trigger the early warning mechanism.

[0119] Based on the Q-learning reinforcement learning algorithm, if there is an optimal damping coefficient mapping relationship in the historical driving data, the load distribution coefficient matrix will be input, and the initial damping sequence adapted to the current road conditions will be output, such as the [front soft / rear hard] mode. This application suppresses the noise of the tire slip rate based on the Kalman filter, and calculates its time derivative as the basis for dynamic adjustment. If the gradient value exceeds the threshold, it indicates a rapid attenuation trend of the tire grip. By integrating the road texture analysis results of the LiDAR point cloud and the suspension travel change rate, a road complexity index is constructed. If the index exceeds the threshold for 3 consecutive frames, it will be determined that a complex road condition has been entered.

[0120] On complex roads, the frequency response bandwidth of the control system is nonlinearly expanded according to the slip rate gradient: High gradient scenario: The bandwidth is increased to 150% of the baseline value to enhance the response speed of damping adjustment;

[0121] In low-gradient scenarios, the bandwidth will be compressed to 80% to suppress the risk of control oscillation; at the same time, the torque distribution feature extraction module dynamically adjusts the motor output phase based on the bandwidth change to achieve cross-domain coordination of damping control and power distribution.

[0122] Introduced in the scenario of continuous curves on gravel roads: the IMU detects that the peak value of the vertical acceleration of the left front wheel reaches a dangerous value, the load distribution coefficient matrix shows that the load proportion of the left front wheel exceeds 45%, and the reinforcement learning model outputs an initial sequence of a 30% increase in the damping of the left front wheel. The slip gradient of the right rear wheel continues to be negative, indicating that the grip of this wheel decreases rapidly. The road complexity index reaches a peak value, and the system expands the control bandwidth to 120%. The torque distribution weight of the right rear wheel is adjusted to 65% simultaneously. The damping of the left front suspension is enhanced to suppress body roll, and the torque of the right rear wheel is increased to compensate for the loss of grip, so that the vehicle passes the curve in a stable posture.

[0123] Embodiment 8:

[0124] This application proposes a system control system for dynamic stability and anti-saturation control, see Figure 8 , based on the real-time offset (ΔCM) of the vehicle's center of mass projection and the yaw rate deviation (Δγ), a two-dimensional state space is constructed. This application uses the Lyapunov function to calculate two dynamic boundaries of the acceleration safety threshold:

[0125] The longitudinal boundary is mapped from the center of mass offset to the front and rear axle torque distribution limit;

[0126] The lateral boundary is the torque difference threshold between the inner and outer wheels determined by the yaw rate deviation;

[0127] For example, if the vehicle changes lanes urgently, causing ΔCM to deviate to the right, the right boundary of the feasible domain will automatically shrink to limit the right wheel torque distribution weight.

[0128] This application uses the exponential approach law to eliminate the chattering phenomenon of traditional sliding mode control. By designing the sliding surface function, the convergence of the control law will be verified based on the Lyapunov stability criterion, so that the reference torque distribution matrix is ​​always within the feasible domain. When the state trajectory is detected to be close to the boundary, the enhanced convergence mechanism is triggered to force the torque distribution to shrink to the center of the feasible domain. The saturation risk of the actuator is determined by real-time monitoring of the brake master cylinder pressure gradient:

[0129] A sudden increase in pressure gradient: This indicates that the brake system is about to reach its hydraulic limit.

[0130] Pressure oscillation frequency exceeds the standard: indicates pressure fluctuation caused by ABS intervention

[0131] Anti-saturation instruction generation:

[0132] This application adopts a dynamic weight distribution method to transfer the excess torque of the saturation risk wheel to the non-saturated wheel in proportion, while compensating for the yaw stability loss.

[0133] For example, when the left front wheel brake pressure approaches the limit, 30% of its torque is dynamically transferred to the right rear wheel, and the heading angle is maintained stable by increasing the right rear wheel steering assist.

[0134] In an actual test, this application combined the scenario of continuous curves on slippery roads, and the vehicle had understeering and brake saturation risks due to sudden changes in the road surface:

[0135] The center of mass projection deviates to the left by 5%, the yaw rate deviation reaches the dangerous threshold, and the left boundary of the dynamic feasible domain shrinks by 20%. The algorithm generates a reference distribution matrix with 60% of the torque to the left front wheel and 40% to the right rear wheel to suppress the understeer tendency.

[0136] The left front brake pressure gradient suddenly increased, and the system identified it as a saturation risk and initiated a redistribution strategy.

[0137] The torque of the left front wheel is reduced to 45%, and that of the right rear wheel is increased to 55%, and the braking intervention of the Electronic Stability Program (ESP) on the right rear wheel is enhanced synchronously. The vehicle resumes its expected trajectory and avoids skidding out of control.

[0138] Embodiment 9:

[0139] This application proposes how to implement a nonlinear constraint collaborative control system, see Fig. 9 The present application first determines whether the total acceleration exceeds the limit according to the objective function. Then, when the limit is exceeded, the present application constructs a distributed optimization model based on the Nash equilibrium theory. Different objective functions reach an equilibrium solution through the interaction of virtual gamers. The competition intensity of the gamers is adjusted by introducing a dynamic relaxation factor. For example, the game weight of the stability target is increased to 70% under emergency conditions.

[0140] This application constructs a dynamic ratio function of the four-wheel vertical force based on the suspension displacement sensor data, and describes the nonlinear relationship between the front / rear axle torque distribution weight and the load ratio through an inverse proportional function.

[0141] For example, when the rear axle torque weight increases, the front wheel load ratio decays exponentially, suppressing the "head-up" effect. In implementation, by adopting the proximal policy optimization (PPO) algorithm, the game objective and load ratio constraint are optimized synchronously in the torque distribution decision. If the nonlinear relationship is detected to deviate from the preset envelope, the reward function penalty mechanism will be triggered to force the strategy to return to the safe area.

[0142] In a test of this application, based on the emergency overtaking scenario on the highway, if the vehicle needs to complete a rapid lane change on a slippery road, the longitudinal acceleration will be controlled to break through the safety threshold, and the vehicle system will activate the multi-objective game process, with the stability target weight increased to 65%, the energy efficiency target reduced to 20%, and the response speed maintained at 15%. The rear axle torque weight is increased to 60%, and the front axle is reduced to 40%, and the electronic stability system (ESP) is synchronously enhanced to intervene in the braking of the inner wheels. The front wheel load ratio is reduced to 35%, and the rear wheel is increased to 65%. The body pitch angle is controlled within 2° to avoid oversteering due to the rearward shift of the center of gravity.

[0143] Embodiment 10:

[0144] This application proposes a new energy vehicle operation optimization system based on intelligent networking, see Fig.10 Specific components include: the road data acquisition module integrates multi-source sensor data such as lidar, millimeter-wave radar, and inertial measurement unit (IMU). This application eliminates the timestamp deviation and spatial coordinate system difference of heterogeneous data through the spatiotemporal alignment algorithm. For example, after the lidar point cloud is fused with the visual semantic data, the micro-texture features of the road ahead can be accurately extracted, such as potholes, cracks, and the curvature and slope of macro-geometric parameters. By configuring the preview mechanism, the road feature data of 500 meters ahead is obtained based on V2X communication, and a dynamic preview window is constructed in combination with a high-precision map.

[0145] If the vehicle is about to enter an unstructured road, the vehicle system will start the data preprocessing process 200 meters in advance to reserve decision time for subsequent modules. The load-acceleration joint modeling module, the load mean model, adopts a hybrid observer architecture to simultaneously analyze the energy exchange process between the sprung mass (body) and the unsprung mass (suspension, wheels). The model parameters are dynamically corrected through the Lyapunov stability criterion to ensure that the output value is always within the physically feasible domain. For example, in the scenario of continuous speed bumps, the model can predict the phase difference distribution of the vertical load of each wheel, providing prior knowledge for suspension control.

[0146] When setting the acceleration safety threshold:

[0147] By constructing a three-dimensional acceleration envelope model, the longitudinal, lateral and vertical acceleration thresholds are dynamically adjusted according to the road adhesion coefficient and the vehicle center of mass position. When a sudden change in tire slip rate is detected, the lateral acceleration allowable range is automatically reduced to suppress the risk of rollover.

[0148] When the suspension damping is adjusted dynamically:

[0149] Based on the predicted output of the load mean model, the fuzzy PID algorithm is used to generate a nonlinear damping curve. Under cornering conditions, the outer suspension damping is enhanced in stages, and the inner damping is released in stages, forming a "hard outside and soft inside" dynamic stiffness distribution to balance the controllability and comfort requirements.

[0150] When torque vectoring is intelligently distributed:

[0151] The sliding mode variable structure control algorithm is used to solve the optimal torque distribution solution within the constraint boundary of the acceleration safety threshold. The anti-saturation compensation mechanism is introduced. When the temperature of a motor approaches the limit, its torque load is automatically transferred to the low-temperature unit, and the power output balance is maintained through ESP braking intervention.

[0152] Introduced into complex road conditions where continuous curves and damaged roads alternate in mountainous areas:

[0153] The lidar detected irregular potholes on the road 50 meters ahead, and the IMU monitored that the current yaw angular velocity deviation exceeded the limit.

[0154] The load model predicts that the vertical load on the right front wheel will suddenly increase by 40%, generating an optimization instruction of "enhanced damping of the right front suspension + increased torque of the left rear wheel". The acceleration safety module lowers the lateral acceleration threshold by 25% due to the slippery road surface (μ = 0.3);

[0155] The suspension module increases the damping of the right front shock absorber to the highest level to suppress impact transmission;

[0156] The torque module distributes 65% of the torque to the left rear wheel and uses the electronic differential effect to compensate for the understeer tendency. The active suspension actuator completes the damping adjustment within 10ms, the vertical acceleration of the vehicle body is reduced by 60%, and the drive motor cooperates with ESP to implement pulse braking on the right front wheel. The yaw angular velocity deviation converges to a safe range.

[0157] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A method for optimizing the operation of new energy vehicles based on intelligent networking, characterized in that: include: The first operating parameter of the vehicle at the current moment and the road parameter at the next moment are collected through the vehicle-mounted sensor network; wherein the first operating parameter includes the tire movement value corresponding to the vehicle suspension deformation variable data under the current road surface coefficient, and the road parameter is the road characteristic data of the road to be entered; Constructing a load mean value model and an acceleration safety threshold value of the current vehicle on the road to be entered according to the first operating parameter and the road characteristic data; Generate a damping adjustment instruction for the vehicle suspension according to the load mean model; Generate a drive torque vector distribution plan based on the acceleration safety threshold.

2. A method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The collecting of the first operating parameter of the vehicle at the current moment and the road parameter at the next moment includes: Obtaining a three-dimensional road excitation response signal collected by a wheel hub embedded MEMS array in a vehicle-mounted sensor under a first operating parameter of the vehicle at the current moment, performing three-dimensional road excitation response signal decomposition, and determining a suspension dynamic deformation characteristic quantity; Through real-time matching of the on-board LiDAR point cloud data in the on-board sensor with the high-precision map in the cloud, the equivalent curvature radius and slope change gradient of the road within the distance to be optimized ahead are calculated to determine the road parameters at the next moment.

3. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The tire movement value collection process includes: Perform frequency domain analysis on the suspension dynamic deformation to extract the target energy distribution in the preset characteristic frequency band associated with the tire resonance frequency; Generate a dynamic compensation factor based on the discrete characteristic parameter of the road ahead; wherein the dynamic compensation factor includes a weighted combination of the road curvature and the slope change rate; Nonlinearly mapping the target energy distribution with the dynamic compensation factor to generate a correction value for the tire contact patch offset; The historical trajectory data of adjacent vehicles is obtained through on-board V2X communication, and the correction value is calibrated in the rolling time domain to generate a tire movement value that integrates multi-source information.

4. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The load mean model is generated based on a hybrid observer model including energy exchange between sprung mass and unsprung mass; wherein the hybrid observer model performs online correction under road characteristic data and performs parameter adaptation according to the online correction.

5. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The constructing of the load mean value model and the acceleration safety threshold value of the current vehicle on the road to be entered also includes: When the predicted value of the load mean exceeds the standard load range, the multi-objective optimization algorithm is started to calculate the optimal stiffness sequence of the damper; A fuzzy PID controller is used to generate a nonlinear adjustment curve of the damping valve opening; wherein the nonlinear adjustment curve includes a dynamic feedforward compensation term based on the slope change rate.

6. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The constructing of the load mean value model of the current vehicle on the road to be entered also includes: Pre-configure the motor temperature constraints of current new energy vehicles and associate them with the battery SOC state to form a feasible domain for torque distribution; Based on the model predictive control framework, the optimal torque distribution solution that satisfies the tire slip ratio constraint is solved in the torque distribution feasible domain.

7. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The generating of the damping adjustment instruction of the vehicle suspension according to the load mean value model comprises: Obtain the vertical acceleration of the sprung mass of the current vehicle and construct a dynamic load distribution coefficient matrix; According to the dynamic load distribution coefficient matrix, under the damping reference prediction, an initial damping coefficient sequence is generated, and a tire slip rate change gradient is obtained; When it is detected that the vehicle is on a preset complex road, the motor torque distribution characteristics are determined based on the tire slip rate change gradient, and the response bandwidth of the damping correction command is dynamically adjusted.

8. The method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 1, characterized in that: The step of generating a driving torque vector distribution scheme according to the acceleration safety threshold comprises: Collect the center of mass projection offset and yaw rate deviation of the current vehicle in real time to generate dynamic feasible domain boundary conditions of the acceleration safety threshold; According to the sliding mode variable structure control algorithm, solving the reference torque distribution matrix that satisfies the Lyapunov stability condition in the dynamic feasible domain; The actuator saturation risk is identified through the brake system pressure feedback data, and a torque redistribution instruction with anti-saturation characteristics is generated to generate a drive torque vector distribution plan.

9. A method for optimizing the operation of new energy vehicles based on intelligent networking as claimed in claim 8, characterized in that: The generating of the driving torque vector distribution scheme according to the acceleration safety threshold also includes: When the actual value of the longitudinal acceleration exceeds the safety threshold, the current vehicle torque distribution coefficient based on the multi-objective game algorithm is activated, and the front / rear axle torque distribution weight coefficient is dynamically adjusted. The four-wheel load ratio output by the load mean prediction model is nonlinearly negatively correlated.

10. A new energy vehicle operation optimization system based on intelligent networking, characterized in that: The system comprises: Road data collection module: collects the first operating parameter of the vehicle at the current moment and the road parameter at the next moment through the vehicle-mounted sensor network; wherein the first operating parameter includes the tire movement value corresponding to the vehicle suspension deformation variable data under the current road surface coefficient, and the road parameter is the road characteristic data of the road to be entered; Load acceleration generation model: used to construct a load mean value model and acceleration safety threshold of the current vehicle on the road to be entered according to the first operating parameter and the road characteristic data; Vehicle suspension adjustment module: used to generate damping adjustment instructions for vehicle suspension according to the load mean model; Torque vector distribution module: used to generate a drive torque vector distribution plan based on the acceleration safety threshold.

Images