Vehicle driving optimization method based on road surface friction coefficient intelligent prediction model

Abstract

The invention discloses a vehicle driving optimization method based on a road surface friction coefficient intelligent prediction model, and belongs to the technical field of urban traffic road engineering and vehicle control. Collecting real-time state parameters of a front road surface in real time through a vehicle vision system, wherein the state parameters comprise road surface types and road surface conditions; according to the real-time state parameters of the front road surface, the running state parameters of the vehicle and the environment state parameters, a real-time prediction friction coefficient value between the vehicle and the road surface is output through the intelligent friction coefficient prediction model; and transmitting the real-time predicted friction coefficient value to a vehicle control system, and adjusting the power distribution of the vehicle. The road friction coefficient value can be predicted in real time, and the power distribution of the vehicle is reasonably adjusted in combination with the current state of the vehicle. A power source of the vehicle is reasonably used, so that the trafficability, the driving economy and the running stability of the vehicle are improved, the driving mileage of the vehicle is prolonged, and the switching of the driving modes of the vehicle is guaranteed to have a basis.

Description

Technical Field

[0001] The present invention belongs to the field of urban traffic road engineering and vehicle control technology, and specifically relates to a vehicle driving optimization method based on an intelligent prediction model of road friction coefficient. Background Art

[0002] With the increase in traffic volume and vehicle speed on my country's high-grade highways and urban roads, the requirements for pavement skid resistance are also increasing. Highway construction and maintenance management departments pay more and more attention to the impact of pavement skid resistance on traffic safety. At present, the testing and evaluation of pavement skid resistance in China mainly adopt the following methods: sand laying method, pendulum friction instrument method, DF tester method, etc., but these methods use manual detection, with low detection efficiency, and the structural depth is only an indirect indicator, which cannot directly reflect the actual friction performance. In addition, although the existing laser scanning structural depth method and lateral force coefficient tester method have high automation and high detection efficiency, the test results are easily affected by external conditions such as water accumulation, snow accumulation, mud, etc., and need to be tested in closed traffic or low traffic volume, and the real-time performance is poor. In addition, the established distance method and locked wheel trailer method are often used to test the friction between the vehicle tire and the road surface after the vehicle is braked and the wheel is locked. The automation level and efficiency are high, but it can only be carried out in an annual inspection manner, with a long window period, and the working condition scene is not enough to reflect the real-time performance of the actual driving environment.

[0003] Because traditional manual detection methods have low detection efficiency and need to be tested in specific environments, the detection results are delayed, the application is not real-time enough, the working scenarios are limited, and they are easily affected by the external environment. Therefore, they cannot meet the high flow and high speed requirements of modern transportation.

[0004] CN119000520A discloses a longitudinal road friction coefficient measurement system, which is mainly used to measure and evaluate the anti-skid performance of highways. The system uses a vehicle-carrying platform to drive the measuring wheel to rotate along the driving direction, and applies additional resistance to the measuring wheel through a braking system, so that the measuring wheel produces a certain amount of sliding when rolling on the road surface. The system calculates the original road friction coefficient by obtaining the vertical load of the measuring wheel, the horizontal resistance of the forward movement, the sliding speed, and the friction between the tire and the road surface. In addition, the system also obtains the vibration acceleration information of the measuring wheel in the vertical road surface direction through a vertical load loading system and an IMU module, and corrects the original road friction coefficient, thereby improving the measurement accuracy. It mainly improves the existing measurement method by designing vertical load loading, vibration acceleration collection, friction coefficient correction, automatic adjustment of watering amount and other systems, but there are still difficulties such as complex system, high implementation difficulty, difficult data processing, and adaptability to be optimized, and it has not been further applied to the prediction of the real-time driving process of the vehicle and the change of the vehicle running state.

[0005] CN118977714A discloses a method for switching driving modes, which intelligently determines and switches the target driving mode (economic mode, sports mode, off-road mode or slippery road mode) of the vehicle by integrating multiple sensor data and environmental detection results, but does not take the predicted road friction coefficient into consideration in the driving mode switching conditions and rationally allocate and optimize the vehicle power accordingly.

[0006] CN115081927A discloses a road friction coefficient evaluation and prediction method, which provides a real-time road friction coefficient warning for road vehicles and provides a medium- and long-term solution, but does not predict the vehicle road friction coefficient in the future extreme time and does not reasonably apply the predicted value to achieve the effect of enhancing the vehicle's operating performance. Summary of the invention

[0007] The purpose of the present invention is to solve the problem that the friction coefficient between the tire and the ground cannot be measured in real time during the vehicle's driving process and fed back to the vehicle to adjust the vehicle's operating state parameters. A vehicle driving optimization method based on an intelligent prediction model of the road friction coefficient is proposed. The road friction coefficient value is predicted in real time, and the power distribution of the vehicle is reasonably adjusted in combination with the current state of the vehicle.

[0008] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0009] A vehicle driving optimization method based on an intelligent prediction model of road friction coefficient comprises:

[0010] The vehicle vision system collects the real-time state parameters of the road ahead, including the road type and road condition. The distance range of the specified distance segment ΔY of the road ahead collected by the vehicle relative to the vehicle is:

[0011]

[0012] Among them, π1(t) is the vehicle speed in the geodetic coordinate system collected at the current moment, TTC min 、TTC max Indicates the upper and lower limits of the collision time, TTC min 、TTC max Indicates the minimum and maximum values ​​of the distance range;

[0013] According to the real-time state parameters of the road ahead, the vehicle's own operating state parameters, and the environmental state parameters, the friction coefficient intelligent prediction model is used to output the real-time predicted friction coefficient value between the vehicle and the road;

[0014] The real-time predicted friction coefficient value is transmitted to the vehicle control system to adjust the vehicle's power distribution.

[0015] The present invention has the following beneficial effects:

[0016] The present invention is based on vehicle engineering technologies such as vehicle dynamics model and tire model and combines computer vision, deep learning and neural network and other computer technologies. By designing vehicle vision and sensor system, the type (material and gradation, texture characteristics, etc.) and condition (wear degree, medium between tire and road surface, etc.) of the road surface in front are collected in real time, and the collected data is transmitted to the big data processing and intelligent prediction module via the vehicle network. In the intelligent prediction model of road friction coefficient designed by the big data processing and intelligent prediction module, the input of the model is the type and condition of the road surface in front collected by the vehicle vision and sensor system, the vehicle's own operating state parameters (driving speed, driving acceleration, side slip angle, slip rate, tire type and wear degree, tire inflation pressure and load, etc.) and environmental state parameters (humidity, temperature, wind resistance, etc.), and the output is the real-time predicted friction coefficient value between the vehicle and the road surface. The value is transmitted to the vehicle control system, and then the vehicle control system reasonably adjusts the power distribution of the vehicle through the predicted value to reasonably use the vehicle power source, thereby improving the vehicle's passability, driving economy, and operating stability, extending the vehicle's mileage and ensuring that the vehicle's driving mode switching is based on evidence.

[0017] Instruction Manual

[0018] Figure 1 The overall architecture diagram of the intelligent prediction, calculation and execution system of the road friction coefficient shown in the embodiment of the present invention

[0019] Figure 2 A diagram showing the operation architecture of a vehicle vision system according to an embodiment of the present invention

[0020] Figure 3 The overall architecture diagram of the big data processing and intelligent prediction module shown in the embodiment of the present invention

[0021] Figure 4 A flow chart showing the optimization of vehicle power distribution according to an embodiment of the present invention DETAILED DESCRIPTION

[0022] The present invention will be further described below in conjunction with specific embodiments. It should be understood that these embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention. In addition, it should be understood that after reading the content taught by the present invention, those skilled in the art can make various changes or modifications to the present invention, and these equivalent forms fall within the scope limited by the appended claims of the application equally.

[0023] The present invention is based on vehicle engineering technologies such as vehicle dynamics model and tire model and combines computer vision, deep learning and neural network and other computer technologies to propose a vehicle driving optimization method based on intelligent prediction model of road friction coefficient. The present invention mainly involves four modules: vehicle vision and sensor acquisition system module, big data processing and intelligent prediction module, vehicle control center system module, and instruction execution VCU module. Its overall architecture is shown in the attached figure. Figure 1 shown.

[0024] First, the vehicle obtains the parameters of the road surface type (material and grading, texture characteristics, etc.) and condition (degree of wear, medium between tire and road surface, etc.) at the specified position in front of the vehicle through the vehicle vision and sensor acquisition system module.

[0025] The designated position is a preset value waypoint, the value of which depends on the preset collision time TTC between the vehicle and the waypoint. Assume that the time it takes for the vehicle processor to calculate an intelligent prediction value of the road friction coefficient and issue an optimized vehicle power adjustment instruction is Δt1, and the preset safety prediction constraint interval is Ω = [Δt min ,Δt max ], then the specific constraint method of collision time is:

[0026] TTC min ≤Δt1+Δt min ≤TTC max ≤Δt1+Δt max

[0027] Among them, Δt min Reserve the minimum time for the preset safety prediction constraint, Δt max Reserve the maximum time for the preset safety prediction constraint, TTC nib 、TTC max Indicates the upper and lower limits of the collision time.

[0028] Then, the distance range of the specified distance segment ΔY of the front road segment value point collected by the vehicle relative to the vehicle is:

[0029]

[0030] Among them, π1(t) is the driving speed in the earth coordinate system collected at the current moment, which can be obtained by converting the speed v(t) in the vehicle coordinate system. To further facilitate control, the present invention converts the coordinates in the vehicle coordinate system and the earth coordinate system:

[0031]

[0032] Among them, x and y represent the longitudinal coordinate and lateral coordinate of the vehicle in the vehicle coordinate system respectively. It is the angle between the line connecting the origins of the two coordinate systems and the horizontal axis of the geodetic coordinate system.

[0033] Furthermore, the vehicle motion related parameters can be expressed as:

[0034]

[0035] Where w represents the wheel rolling angular velocity, and r represents the wheel radius. π2(t) represents the vehicle acceleration in the geodetic coordinate system collected at the current moment; π3(t) represents the vehicle side slip angle collected at the current moment; π4(t) represents the vehicle slip rate collected at the current moment; π5(t) represents the vehicle tire type and wear degree collected at the current moment, which is calculated based on the vehicle tire type τ type , tire wear rate τ wear and remaining useful life τ life Perform fuzzy mathematical evaluation on the actual mileage and time, and calculate the evaluation score between [0,1] as the return value, F fce (τ type ,τ wear ,τ life ) represents the fuzzy mathematical evaluation process; π6(t) represents the vehicle tire inflation pressure and load collected at the current moment; π7(t) represents the weighted value of other parameter variables that indirectly affect the vehicle's own operating state collected at the current moment, and the default value is 0. When the vehicle detects other parameter variables that indirectly affect the vehicle's own operating state, all parameter variables that affect the vehicle's own operating state are fuzzily evaluated in [0,1) and the weighted value is taken as its value at that moment; τ Pressure With τ Load They refer to the abnormal index of vehicle tire pressure and vehicle load, respectively, and the abnormal index of vehicle tire pressure τ Pressure It is the percentage of the vehicle tire real-time pressure deviating from the standard pressure. The specific calculation method is:

[0036]

[0037] Where P(t) represents the pressure of the vehicle tire at the current moment, P min Indicates the lowest pressure of the tire under normal conditions, P max Indicates the maximum pressure of the tire under normal conditions. Abnormal index of vehicle load τ Load The calculation method is similar:

[0038]

[0039] Where L(t) represents the load of the vehicle at the current moment, L max Indicates the rated maximum load of the vehicle.

[0040] From this, we can get the parameter matrix of the vehicle's own operating state parameters:

[0041] Π(t)=[π1(t),π2(t),π3(t),π4(t),π5(t),π6(t),π7(t)]

[0042] like Figure 2 The vehicle vision system operation architecture diagram shown in the figure mainly includes two cameras, the main camera and the auxiliary camera. The main steps of obtaining the real-time state parameter matrix of the road surface are as follows:

[0043] Step 1: The vehicle visual system obtains the target road section image captured by the vehicle main camera at high speed, wherein the image is the road condition image captured when the vehicle reaches the preset distance ΔY.

[0044] Step 2: Input the target road section image into a pre-trained convolutional neural network for feature recognition to obtain the pre-recognition type and overall recognition probability of the target road surface feature, wherein the attributes of the target road surface feature should include but not be limited to road surface type (material and gradation, texture characteristics, etc.) and condition (wear degree, medium between tire and road surface, etc.) parameters. Here, the pre-trained convolutional neural network can be implemented based on existing technologies.

[0045] Step 3: If the overall recognition probability of the target road surface is greater than or equal to the preset probability threshold c, skip step 4 and execute step 5.

[0046] Step 4: If the overall recognition probability of the target road surface is lower than the preset road condition probability threshold c, the road surface image captured by the vehicle's secondary camera is obtained for secondary judgment, where the road surface image is a combination of road surface image sequences continuously captured by the secondary camera within a preset time period.

[0047] If the judgment result is the same as the first judgment result, the frame is marked as empty and a road surface missing report is sent. At the same time, the road surface image is manually marked and stored in the database for periodic updating of the convolutional neural network. Return to step 1 and retake a new road surface image.

[0048] Otherwise, the target road surface photographed by the vehicle's secondary camera is used to execute step five.

[0049] Step 5: Convert the characteristic attributes of the target road surface into a real-time road surface state parameter matrix and send it to the big data processing and intelligent prediction module.

[0050] The road surface real-time state parameter matrix Θ(t) is expressed as:

[0051] Θ(t)=[θ1(t),θ2(t),θ3(t),θ4(t),θ5(t)]

[0052] Among them, θ1(t) represents the pavement material and grading parameters collected at the current moment, θ2(t) represents the pavement texture characteristic parameters collected at the current moment, θ3(t) represents the wear degree level collected at the current moment, θ4(t) represents the medium type parameter between the tire and the pavement collected at the current moment, and θ5(t) represents the weighted value of other parameter variables that indirectly affect the real-time state of the pavement collected at the current moment.

[0053] Synchronously, the real-time state parameter matrix E(t) of the environment can be obtained through the environmental sensors carried by the vehicle itself, which should be:

[0054] E(t)=[ε1(t),ε2(t),ε3(t),ε4(t)]

[0055] Among them, ε1(t) represents the ambient humidity value collected at the current moment, ε2(t) represents the ambient temperature value collected at the current moment, ε3(t) represents the ambient wind resistance value collected at the current moment, and ε4(t) represents the weighted value of other environmental variables that indirectly affect the vehicle's driving resistance collected at the current moment.

[0056] Then, if Figure 3 As shown, the vehicle big data processing and intelligent prediction module receives the vehicle's own operating state parameter matrix Π(t), the road surface real-time state parameter matrix Θ(t), and the environment real-time state parameter matrix E(t) from the same timestamp, and inputs them into the road friction coefficient intelligent prediction model to obtain feedback of the road friction coefficient prediction value μ(t+1).

[0057] The construction method of the road friction coefficient intelligent prediction model is as follows:

[0058] Step 1: Obtain road friction coefficient training samples under different driving conditions to build a sample database. Each sample data contains all state parameters of the vehicle's own operating state parameter matrix Π(t), the road real-time state parameter matrix Θ(t), and the environment real-time state parameter matrix E(t), a total of 16 components;

[0059] Step 2: Establish an intelligent prediction model of road friction coefficient based on BP neural network and initialize the network. The network initialization includes setting the maximum number of training times N, the number of hidden nodes m, the threshold c, the initial weight w, the initial learning rate v, the expected learning accuracy p and other parameters.

[0060] Step 3: Training sample input.

[0061] The training sample input is an n×16 training sample matrix:

[0062]

[0063] Step 4: Establish a BP neural network for intelligent prediction of road friction coefficient for driving samples. First, evenly distribute the training samples to each Map node. Then, read the network weight record from the Hadoop distributed file system by calling the Map function. Each Map task will obtain and output the corresponding weight. Next, the Reduce function reads the network weight record from the Hadoop distributed file system and receives the weight output by the Map task. The Reduce task determines whether the next iteration is needed by comparing the difference between the updated network weight of the Map node and the weight stored in the Hadoop file system.

[0064] Step 5: If the network weights are no longer updated, the BP neural network training of the road friction coefficient intelligent prediction model is completed and the construction is preliminarily completed.

[0065] Step 6: Set up a priori model data package supplement library. During operation, if new road surface types or other variable parameters are found, they can be further added to the library. At the same time, continuous optimization and correction are performed based on the comparison between actual operation data and the true value of big data.

[0066] like Figure 4 As shown in the figure, after obtaining the predicted value of the road friction coefficient μ(t+1), the vehicle control center system module can adjust the vehicle power distribution and optimize the vehicle driving. Specifically, the vehicle control center system module obtains the predicted value of the road friction coefficient, and the vehicle control center system module determines the target driving mode of the vehicle based on the vehicle speed, vehicle suspension state, accelerator pedal and brake pedal state, and wheel speed of each wheel, combined with the road friction, and outputs it to the command execution VCU module, and the command execution VCU module controls the vehicle driving mode to switch to the target driving mode.

[0067] The specific power distribution strategy and driving mode switching design and algorithm design are as follows:

[0068] The core of the power distribution strategy is to reasonably adjust the driving force, braking force and lateral force of the front and rear axles based on the real-time predicted road friction coefficient value and the current state of the vehicle.

[0069] For the driving force / braking force distribution of the front and rear axles, the total driving force is obtained according to the product of the current acceleration / deceleration and the vehicle weight, and the maximum available driving force of the front and rear axles is calculated according to the front and rear axle load ratio and the predicted value of the friction coefficient; if the total driving force does not exceed the sum of the maximum available driving forces of the front and rear axles, the total driving force is distributed according to the front and rear axle load ratio, otherwise the maximum available driving force of the front and rear axles is used as the driving force / braking force after distribution, and the total driving force is reduced by reducing the engine power, using the braking force or adjusting the pressure of the brake system;

[0070] For lateral force distribution, the total lateral force is calculated based on the current driving speed, sideslip angle, vehicle wheelbase and vehicle weight, and the maximum available lateral force of the front and rear axles is calculated based on the front and rear axle load ratio and the predicted value of the friction coefficient; if the total lateral force does not exceed the sum of the maximum available lateral forces of the front and rear axles, the total lateral force is distributed according to the front and rear axle load ratio, otherwise the maximum available lateral force of the front and rear axles is used as the distributed lateral force, and the total lateral force is reduced by adjusting the steering angle or speed of the vehicle.

[0071] The specific steps are as follows:

[0072] For driving force distribution: Input data include the predicted road friction coefficient μ pred , current driving speed v, current acceleration a, vehicle weight m, load ratio of front and rear axles: k f and k r (Front axle load ratio k f Ratio to rear axle load k r and 1), medium parameters between tire and road surface: such as water film thickness, mud and sand coverage, etc.

[0073] Calculate the total driving force:

[0074] F drive_total =m·a

[0075] Calculate the maximum available driving force on the front and rear axles:

[0076] F drive_f_max =μ pred ·k f ·m·g

[0077] F drive_r_max =μ pred ·k r ·m·g

[0078] Where g is the acceleration due to gravity (about 9.81 m / s 2 ).

[0079] Determine the distribution of drive force between the front and rear axles: Compare the calculated total drive force F drive_total Maximum available driving force F of the front and rear axles drive_f_max and F drive_r_max , if F drive_total ≤F drive_f_max +F drive_r_max , the driving force can be distributed proportionally:

[0080] F drive_f =k f ·F drive_total

[0081] F drive_r =k r ·Fdrive_total

[0082] If F drive_total >F drive_f_max +F drive_r_max , then measures need to be taken to reduce the total driving force and ensure safety:

[0083] F drive_f =F drive_f_max

[0084] F drive_r =F drive_r_max

[0085] F drive_reduce =F drive_total -(F drive_f_max +F drive_r_max )

[0086] F is achieved by reducing engine power or using braking force drive_reduce reduction.

[0087] For braking force distribution: Input data include the predicted road friction coefficient μ pred , current driving speed v, current deceleration a b , vehicle weight m, load ratio of front and rear axles k f and k r ;

[0088] Calculate the total driving force:

[0089] F brake_total =m·a b

[0090] Calculate the maximum available driving force on the front and rear axles:

[0091] F brake_f_max =μ pred ·k f ·m·g

[0092] F brake_r_max =μ pred ·k r ·m·g

[0093] Determine the distribution of drive force between the front and rear axles: Compare the calculated total drive force F brake_total Maximum available driving force F of the front and rear axles brake_f_max and F brake_r_max , if F brake_total ≤F brake_f_max +F brake_r_max , the driving force can be distributed proportionally:

[0094] F brake_f =k f ·Fbrake_total

[0095] F nrake_r =k r ·F brake_total

[0096] If F brake_total >F brake_f_max +F brake_r_max , then measures need to be taken to reduce the total driving force and ensure safety:

[0097] F brake_f =F brake_f_max

[0098] F brake_r =F brake_r_max

[0099] F brake_reduce =F brake_total -(F brake_f_max +F brake_r_max )

[0100] F is achieved by adjusting the pressure of the brake system brake_reduce reduction.

[0101] For lateral force distribution: Input data include the predicted road friction coefficient μ pred , current driving speed v, current sideslip angle δ, vehicle weight and center position m and h, wheelbase t, load ratio of front and rear axles k f and k r ;

[0102] Calculate the total lateral force:

[0103]

[0104] Where L is the wheelbase of the vehicle.

[0105] Calculate the maximum available lateral force on the front and rear axles:

[0106] F lateral_f_mac =μ pred ·k f ·m·g

[0107] F lateral_r_max =μ pred ·k r ·m·g

[0108] Determine the lateral force distribution between the front and rear axles: Compare the calculated total lateral force F lateral_total Maximum available lateral force F on the front and rear axles lateral_f_max and F lateral_r_max , if F lateral_total ≤F lateral_f_max +Flateral_r_max , the lateral force can be distributed proportionally:

[0109] F lateral_f =k f ·F lateral_total

[0110] F lateral_r =k r ·F lateral_total

[0111] If F lateral_total >F lateral_f_max +F lateral_r_max , then measures need to be taken to reduce the total lateral force and ensure safety:

[0112] F lateral_f =F lateral_f_max

[0113] F lateral_r =F lateral_r_max

[0114] F lateral_reduce =F laterak_total -(F lateral_f_max +F lateral_r_max )

[0115] F is achieved by adjusting the vehicle's steering angle or speed lateral_reduce reduction.

[0116] In addition, in a specific implementation of the present invention, it is also necessary to intelligently determine the most suitable driving mode based on the predicted value of the road friction coefficient and the current state of the vehicle, and notify the vehicle control system to adjust the corresponding parameters; when intelligently determining the most suitable driving mode, use the predefined evaluation rules to evaluate the current most suitable driving mode. If the evaluated driving mode is different from the current mode and the switching conditions meet the requirements, then switch; otherwise, do not switch. The driving modes include normal mode, wet mode, snow mode, and muddy mode; each mode has a predefined parameter adjustment scheme.

[0117] In this embodiment, defining the driving mode includes:

[0118] Normal Mode: Suitable for dry, flat roads.

[0119] Wet Mode: Suitable for slippery and wet roads.

[0120] Snow Mode: Suitable for roads covered with ice and snow.

[0121] Mud Mode: Suitable for muddy and soft roads.

[0122] Create evaluation rules:

[0123] Use decision trees or rule engines to evaluate the most suitable driving mode at the moment based on different road friction coefficient values, vehicle status and environmental conditions.

[0124] For example: If μ pred <0.4 and humidity>80%, switch to wetland mode; if μ pred <0.2 and temperature <0℃, and the road type is ice and snow, switch to snow mode; if μ pred <0.3 and mud coverage >50%, switch to muddy mode.

[0125] When making a mode switching decision, the system first evaluates the driving mode in real time, obtains the latest friction coefficient prediction value and other relevant data from the intelligent prediction module at regular intervals (such as 0.1 seconds), and uses predefined evaluation rules to evaluate the most suitable driving mode at the moment.

[0126] Switching conditions, such as a large predicted change in the friction coefficient, high vehicle speed, or a driver request, determine whether the mode switches.

[0127] For example: If you switch from normal mode to wet mode, make sure the vehicle speed is less than 60km / h; if you switch from wet mode to snow mode, make sure the vehicle speed is less than 40km / h.

[0128] The present invention sends a mode switching instruction through a vehicle control center system (such as an ECU or other control unit).

[0129] The instructions include the new driving mode and its corresponding parameter adjustment requirements.

[0130] In a proposed implementation of the present invention, different driving models predefine parameter adjustment schemes for the engine management system (EMS), brake control system (BCS), steering control system (SCS), and suspension system (Suspension System).

[0131] Engine Management System (EMS): adjusts engine output to ensure safety and economy.

[0132] Normal mode: normal power output.

[0133] Wet mode: reduces power output to prevent tire slippage.

[0134] Snow mode: further reduces power output and uses a lower gear.

[0135] Mud Mode: Adjusts torque output, uses low gear, and increases anti-slip control.

[0136] Brake Control System (BCS): Adjusts brake force distribution to ensure balanced braking force on the front and rear axles.

[0137] Normal mode: Standard braking force distribution.

[0138] Wetland mode: increases rear axle braking force and reduces front axle braking force to prevent front wheel slippage.

[0139] Snow mode: Try to keep the braking force of the front and rear axles balanced and increase the intervention of the Electronic Stability Program (ESP).

[0140] Mud mode: increases front axle braking force, reduces rear axle braking force, and increases wheel anti-lock braking system (ABS) intervention.

[0141] Steering Control System (SCS): Adjusts steering angle and steering assistance to optimize vehicle handling.

[0142] Normal mode: Standard power steering.

[0143] Wetland Mode: Moderately increase steering assistance to improve handling stability.

[0144] Snow Mode: Greatly increases steering assist and reduces steering angle changes.

[0145] Mud Mode: Moderately reduces steering assistance and increases the limit of steering angle changes.

[0146] Suspension System: Adjust the damping and stiffness of the suspension to optimize the vehicle's driving stability.

[0147] Normal mode: Standard suspension settings.

[0148] Wet mode: Increases suspension damping and reduces body roll.

[0149] Snow mode: further increases suspension damping and improves driving stability.

[0150] Mud Mode: Reduces suspension damping, increases suspension stiffness, and improves passability.

[0151] After the mode is switched, the vehicle status and road conditions continue to be monitored in real time. If new road conditions change or the vehicle status is abnormal, the parameters can be dynamically adjusted or the mode switch can be triggered again. For example, if the road surface is detected to be dry in wet mode, it can be re-evaluated and switched back to normal mode.

[0152] The above examples are only specific embodiments of the present invention. Obviously, the present invention is not limited to the above examples, and many variations are possible. All variations that can be directly derived or associated with the contents disclosed by a person skilled in the art should be considered as the protection scope of the present invention.

Claims

1. A vehicle driving optimization method based on an intelligent prediction model of road friction coefficient, characterized in that: include: The vehicle vision system collects the real-time status parameters of the road ahead, including the road type and condition; The distance range of the specified distance segment ΔY of the front road segment collected by the vehicle relative to the vehicle is: Among them, π1(t) is the vehicle speed in the geodetic coordinate system collected at the current moment, TTC min 、TTC max Indicates the upper and lower limits of the collision time, TTC min 、TTC max Indicates the minimum and maximum values ​​of the distance range; According to the real-time state parameters of the road ahead, the vehicle's own operating state parameters, and the environmental state parameters, the friction coefficient intelligent prediction model is used to output the real-time predicted friction coefficient value between the vehicle and the road; The real-time predicted friction coefficient value is transmitted to the vehicle control system to adjust the vehicle's power distribution.

2. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The process of collecting the real-time status parameters of the road ahead through the vehicle vision system is as follows: (1) Obtaining the target road section image captured by the vehicle's main camera at high speed through the vehicle's visual system, that is, the road condition image captured when the vehicle reaches a preset distance; (2) inputting the target road section image into a pre-trained convolutional neural network for feature recognition to obtain a pre-recognition type and overall recognition probability of the target road surface feature; (3) If the overall recognition probability of the target road surface is greater than or equal to the preset probability threshold c, directly execute step (4); If the overall recognition probability of the target road surface is lower than a preset road condition probability threshold c, a road surface image captured by the vehicle's secondary camera is obtained for secondary determination, where the road surface image is a combination of road surface image sequences captured continuously by the secondary camera within a preset time period; If the judgment result is the same as the first judgment result, the frame is marked as empty and a road surface missing report is sent. At the same time, the road surface image is manually marked and stored in the database for regular updating of the convolutional neural network. Return to step (1) and retake a new road surface image; otherwise, use the target road surface photographed by the vehicle's secondary camera to execute step (4); (4) Convert the characteristic attributes of the target road surface into a real-time road surface state parameter matrix.

3. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 or 2 is characterized in that: The real-time road surface status parameters at least include road surface material and gradation, texture characteristics, road surface wear degree, and the medium between the tire and the road surface.

4. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The upper and lower limits of the collision time meet the following conditions: TTC min ≤Δt1+Δt min ≤TTC max ≤Δt1+Δt max Among them, Δt min , Δt max are the minimum time and maximum time of the preset safety prediction constraint interval respectively, and Δt1 is the time for real-time prediction of the friction coefficient value and issuing the optimized vehicle power adjustment command.

5. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The vehicle's own operating state parameters include driving speed, driving acceleration, sideslip angle, slip rate, tire type and wear degree, tire inflation pressure and load, and the calculation formula is as follows: Among them, π1(t) represents the driving speed collected at the current moment, They represent the first-order derivatives of the longitudinal coordinate X(t) and the transverse coordinate Y(t) of the vehicle in the geodetic coordinate system, respectively. x and y represent the longitudinal coordinate and transverse coordinate of the vehicle in the vehicle coordinate system, respectively. π2(t) represents the driving acceleration in the geodetic coordinate system collected at the current moment. represents the first-order derivative of π1(t), represents the second-order derivative of Y(t); π3(t) represents the vehicle slip angle collected at the current moment; π4(t) represents the vehicle slip rate collected at the current moment, represents the wheel rolling angular velocity, and r represents the wheel radius; π4(t) represents the vehicle tire type and wear degree collected at the current moment, which is based on the vehicle tire type τ type , tire wear rate τ wear and remaining useful life τ life Perform fuzzy mathematical evaluation on the actual mileage and time, and calculate the evaluation score between [0,1] as the return value, F fce (τ type ,τ wear ,τ life ) represents the fuzzy mathematical evaluation process; π6(t) represents the vehicle tire inflation pressure and load collected at the current moment; π7(t) represents the weighted value of other parameter variables that indirectly affect the vehicle's own operating status collected at the current moment, and the default value is 0.

6. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The environmental state parameters include humidity, temperature and wind resistance.

7. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The friction coefficient intelligent prediction model adopts BP neural network.

8. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: The real-time predicted friction coefficient value is transmitted to the vehicle control system to adjust the power distribution of the vehicle, including the driving force, braking force and lateral force distribution of the front and rear axles; For the driving force / braking force distribution of the front and rear axles, the total driving force is obtained by multiplying the current acceleration / deceleration and the vehicle weight, and the maximum available driving force of the front and rear axles is calculated based on the front and rear axle load ratio and the predicted value of the friction coefficient; If the total driving force does not exceed the sum of the maximum available driving forces of the front and rear axles, the total driving force is distributed according to the load ratio of the front and rear axles. Otherwise, the maximum available driving force of the front and rear axles is used as the distributed driving force / braking force, and the total driving force is reduced by reducing the engine power, using the braking force or adjusting the pressure of the brake system. For lateral force distribution, the total lateral force is calculated based on the current driving speed, sideslip angle, vehicle wheelbase and vehicle weight, and the maximum available lateral force of the front and rear axles is calculated based on the front and rear axle load ratio and the predicted value of the friction coefficient; If the total lateral force does not exceed the sum of the maximum available lateral forces of the front and rear axles, the total lateral force will be distributed according to the ratio of the front and rear axle loads. Otherwise, the maximum available lateral forces of the front and rear axles will be used as the distributed lateral force, and the total lateral force will be reduced by adjusting the steering angle or speed of the vehicle.

9. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 1 is characterized in that: Also includes: According to the predicted road friction coefficient and the current state of the vehicle, it intelligently determines the most suitable driving mode and notifies the vehicle control system to adjust the corresponding parameters; When intelligently determining the most suitable driving mode, the predefined evaluation rules are used to evaluate the most suitable driving mode at the moment. If the evaluated driving mode is different from the current mode and the switching conditions meet the requirements, the mode is switched; Otherwise, do not switch.

10. The vehicle driving optimization method based on the road friction coefficient intelligent prediction model according to claim 9 is characterized in that: The driving modes include normal mode, wetland mode, snow mode and muddy mode; each mode is provided with a predefined parameter adjustment scheme.

Images