Torque vector prediction method based on simplified vehicle dynamics model

Abstract

The invention discloses a torque vector prediction method based on a simplified vehicle dynamics model. The torque vector prediction method comprises the following steps: S10, acquiring real-time vehicle body attitude and wheel speed data of a vehicle in a vehicle bus; the real-time motion state and direction of the vehicle and the real-time external acceleration are obtained through a composite sensor; on the basis of the vehicle dynamics model, a simplified vehicle dynamics model is obtained, and the existing path of the vehicle is calculated; s20, obtaining steering input data and pedal input data, and obtaining a driver prediction vector by using the driving input prediction model; and S30, performing difference operation on the calculated vehicle existing path and the driver prediction vector, and obtaining a torque vector output result by using a torque vector calculation model. The predictive torque vector algorithm can be used under relatively wide working conditions, has high adaptability to vehicles with different dynamic characteristics, and can enhance the functions (ABS and TC) of a vehicle passive driving assistance system or directly perform internal integration on the vehicle passive driving assistance system.

Description

Technical Field

[0001] The present invention belongs to the technical field of vehicle detection, and particularly relates to a torque vector prediction method based on a simplified vehicle dynamics model. Background Art

[0002] The present invention relates to vehicle dynamics, vehicle dynamic control, data fitting, and motor control algorithms

[0003] Torque vectoring is a form of chassis control that applies additional differential steering to the inner and outer tires during a vehicle's cornering process through, including but not limited to, motor control, differential control, and brake control. By utilizing the longitudinal tire grip that is generally rarely utilized during mechanical steering, it realizes a turning moment in the same direction as the rotation direction to assist the vehicle in cornering, thereby improving the vehicle's dynamic limit.

[0004] The current vehicle torque vector algorithms can be mainly divided into the following control modes:

[0005] Pure differential torque vectoring based on steering wheel angle deflection and pedal opening, grip torque vectoring based on vehicle acceleration and throttle / brake opening, and the recently increasingly popular AI large model torque vector algorithm.

[0006] ① Differential torque vectoring algorithms (such as BYD Seal, Hyundaiioniq 5N) mainly rely on comprehensively judging pedal inputs and steering wheel angle inputs, and based on these inputs, according to the look-up table (Map) made by chassis engineers in advance, independently control the motor, mechanical brakes, or differentials to enable the vehicle to achieve the ability of torque vectoring. Since the differential torque vectoring algorithm is based on a completely open-loop control mode, although it has the advantage of stable reproducibility, on the one hand, because it does not consider any external control factors (such as whether the vehicle is going uphill or downhill, whether the ground is slippery, etc.), it not only poses relatively harsh requirements on the tuning work itself, but also, due to its open-loop control characteristics, it not only cannot enhance the effectiveness of general passive driving aids (such as ABS, TC, etc.), but on the contrary, because this algorithm consumes additional lateral tire grip, it makes the working window of the passive driving assistance system narrower.

[0007] ②The grip-type torque vectoring algorithm (e.g., BMW G80 / G82 M3 / M4, Volkswagen ID.3) not only relies on driver inputs (i.e., pedal and steering wheel inputs), but also on some sensors located on the vehicle's suspension structure (such as attitude sensors on the body's lower control arm, optical wheel speed sensors) and body acceleration sensors. Based on the vehicle's static dynamics model (i.e., assuming that the four wheels do not slide significantly relative to the ground, or remain within a certain pre-given preset operating condition range) and some pre-programmed look-up tables, it independently controls the motor, mechanical brakes or differential, enabling the vehicle to achieve torque vectoring capabilities. Since the grip-type torque vectoring algorithm is calculated based on a variety of different vehicle sensors and preset algorithms, compared with the former, its control method is more linear and predictable, and its grip efficiency is also higher. However, since it is still based on a completely open-loop control, although it takes into account a large number of possible impacts from the external environment, it still cannot enhance the working window of general passive driving assistance systems. Moreover, since it is based on a vehicle dynamics model under a certain given operating condition, once the vehicle deviates from the preset operating condition, the algorithm will no longer be able to control the vehicle reasonably.

[0008] ③The AI large model torque algorithm highly depends on AI learning. Through various data feedback provided by the test tools on the chassis, the AI large model learns the vehicle dynamics characteristics, thus forming a certain AI or AI + algorithm. Finally, through the deployment of this algorithm on the vehicle controller, it independently controls the motor, mechanical brakes or differential, enabling the vehicle to achieve torque vectoring capabilities. Since the AI large model algorithm is almost entirely based on an AI large model with a very high black box level, although it may have intelligent characteristics that other torque vector models programmed by human engineers based on a certain fixed algorithm do not have, it does not have strong reproducibility and stability. That is to say, for example, when a human driver drives the vehicle aggressively and needs to perform continuous rapid and precise operations on the vehicle, during such an operation process, not only may this algorithm cause accident risks due to misjudging the driver's frequent and irregular inputs, but the uncertainty it shows itself also cannot give enough confidence to the human driver performing this operation. Moreover, the AI large model consumes a huge amount of computing power, and car manufacturers need to collect a large amount of data for this technology again for each new vehicle to ensure that the system can adapt to the new vehicle, which will cause a huge waste of resources. Summary of the Invention

[0009] To solve the above problems, the present invention proposes a torque vector prediction method based on a simplified vehicle dynamics model, which can be used under relatively broad operating conditions, has high adaptability to vehicles with different dynamics characteristics, and can enhance the functions of the vehicle's passive driving assistance systems (ABS, TC), or directly integrate them internally, a predictive torque vectoring algorithm.

[0010] To achieve the above object, the technical solution adopted by the present invention is: a torque vector prediction method based on a simplified vehicle dynamics model, including the steps of:

[0011] S10. Obtain the real-time body attitude and wheel speed data of the vehicle in the vehicle bus; use a composite sensor to obtain the real-time motion state and direction of the vehicle, as well as the real-time external acceleration; based on the vehicle dynamics model, obtain a simplified vehicle dynamics model and calculate the existing path of the vehicle.

[0012] S20. Obtain the steering input data and pedal input data, and use the driving input prediction model to obtain the driver prediction vector.

[0013] S30. After performing a difference operation on the calculated existing path of the vehicle and the driver prediction vector, use the torque vector calculation model to obtain the torque vector output result.

[0014] Furthermore, in the simplified vehicle dynamics model:

[0015] By combining the suspension motion data obtained by the suspension sensor, the real-time geometric position and deformation amount of the inner wall of the wheel, the wheel size, the geometric position of the vehicle center of gravity including two parts: the center of gravity height CGH and the vehicle corner weight, the vehicle acceleration, the spring value of the suspension, the data collected by the vehicle four-wheel speed sensors in the bus, the driver's steering and pedal inputs, two spring-damping models and the equivalent grip benefit value corresponding table of the corresponding tire deformation conditions to form a simplified tire model, the actual motion trajectory of the vehicle, the length, width, and height of the vehicle, as well as the wheelbase and track width, to construct the simplified vehicle dynamics model during the vehicle motion process.

[0016] Furthermore, the construction process of the simplified vehicle dynamics model includes the steps of:

[0017] S101. According to the suspension motion data obtained by the sensor during the vehicle motion, the geometric position of the inner wall of the wheel and the wheel size, and according to the geometric appearance of the suspension, calculate the wheel ground contact midpoint position P_CP, and obtain the velocity instantaneous center P_IC of any motion state of the suspension and the suspension rotation center P_RC.

[0018] S102. By synthesizing the suspension rotation center P_RC, obtain the roll rotation axis of the vehicle at any time; according to the roll rotation axis and the vehicle real-time acceleration, calculate the original angular momentum of the vehicle at any time.

[0019] S103. According to the suspension motion data, substitute the actual transverse and longitudinal moments of inertia into the suspension motion data; since the spring value of the suspension and the geometric position of the suspension system are known, obtain the effective damping value of any motion condition of the suspension and the suspension speed threshold of the fast and slow valve switches.

[0020] S104, constructing a vehicle suspension model: performing real-time force analysis on the vehicle suspension geometry based on the vehicle's real-time yaw and pitch data, thereby deriving the internal stress of the suspension system. Based on the internal stress of the suspension and the deformation tracking of the suspension bushings achieved through visual tracking, independent force analysis is performed on each bushing, and then the equivalent hardness value of the suspension flexible component is calculated based on the deformation under different external forces.

[0021] S105: Using the measured suspension motion data, raw angular momentum, suspension velocity thresholds for fast and slow valve opening and closing, and equivalent stiffness values, all forces acting on the tire at any given time are calculated, including forces caused by the vehicle's center of gravity shift, forces caused by the mutual traction of rigid components within the suspension system, and forces caused by bushings of flexible suspension components.

[0022] S106: Derivative the constructed vehicle suspension model. By combining the force analysis obtained from the suspension geometry, the real-time tire load is obtained. The vehicle accelerations on the x, y, and z axes and the vehicle slip angle derived from the vehicle's motion are then incorporated into the overall vehicle grip force vector to obtain a real-time grip estimate for each wheel. The tire slip ratio κ is then calculated based on the difference between the wheel speed sensor and the calculated ground speed.

[0023] S107, based on the total load W on the tire in the scope Transfer,sum With four-wheel lateral grip F y The corresponding relationship between the wheel deformation and the grip equivalent μ is introduced, where

[0024] μ=F y / W Transfer,sum κ;

[0025] This forms the first part of the simplified tire model, namely the empirical model of tire horizontal grip;

[0026] S107, based on the correspondence between the total load GWT on the scope tire and the four-wheel mechanical grip Fy, derive a grip equivalent Mu corresponding to the wheel deformation: Mu = Fy / GWT x SR, forming the first part of the simplified tire model, i.e., the horizontal grip prediction model;

[0027] S108: Given the known suspension system load and wheel deformation, the entire wheel is considered to be two longitudinal spring-damper structures connected by the tread. Based on the external force exerted by the suspension on the tire and the actual deformation of the tire itself, the equivalent tire load-lateral grip relationship is calculated to form the second part of the simplified tire model, namely, the tire force empirical model.

[0028] Based on the suspension motion data measured in S101, the original angular momentum obtained in S102, the suspension speed threshold for the fast and slow valve switches obtained in S103, the equivalent hardness value obtained in S104, the tire horizontal grip empirical model obtained in S107, and the tire force empirical model obtained in S108, a vehicle dynamics model is comprehensively obtained.

[0029] Furthermore, from the simplified vehicle dynamics model, the existing vehicle path is deduced, including:

[0030] Assume that the vehicle does not use any active power distribution to assist steering and only relies on the lateral grip of the vehicle under the current conditions for steering. Its motion trajectory is calculated by relying on the above-mentioned simplified vehicle dynamics model for the mechanical calculation of the next moment to obtain the existing vehicle path.

[0031] Furthermore, in the driving input prediction model:

[0032] Based on the offset of the absolute value of the steering wheel angle from the current pointing state of the vehicle, the acceleration of the steering wheel angle, the output value of the absolute value of the pedal input corresponding to the preset throttle curve, the acceleration of the pedal input, and the preset longitudinal torque vector offset comparison table, a vector value of the vehicle operation direction expected by the driver himself is comprehensively deduced, that is, the driver prediction vector output.

[0033] Furthermore, in the driving input prediction model:

[0034] When the vehicle discovers through GPS that there is an upcoming sharp turn that requires heavy braking ahead, it will choose to reduce the offset weight of the driver prediction vector D during the rapid deceleration process and increase the weight in the acceleration and deceleration directions of D, so as to reduce the magnitude of the torque vector difference TV and increase the handling stability characteristics during the vehicle deceleration process; similarly, if the system discovers that the vehicle is about to approach a corner, it will increase the offset weight of D and reduce its weight in the acceleration and deceleration directions, so as to reduce the handling stability characteristics under specific conditions in exchange for better turning characteristics.

[0035] Furthermore, the system compares the un-offset D vector with the offset DP vector. If the offset of both is higher than a certain preset value, the system will calculate the torque vector based on the D vector.

[0036] Furthermore, in the torque vector calculation model:

[0037] Ⅰ: The real-time pointing, speed, and acceleration of the entire vehicle, and the differences in pointing, speed, and acceleration between it and the driver prediction vector obtained from the driving input prediction model in ②; Ⅱ: The real-time available grip of the four wheels output by the simplified vehicle dynamics model; Ⅲ: The real-time attitude and rudder angle information of the vehicle;

[0038] Using the data obtained in steps I, II, and III, we will attempt to resolve the difference obtained in step I on each of the four wheels based on the calculated real-time available tire grip, thereby deriving the final specific output method of torque vectoring.

[0039] The beneficial effects of adopting this technical solution are:

[0040] Compared to differential torque vectoring models, the present invention is based on a simplified vehicle dynamics model rather than simply mapping to a specific table. Therefore, the present invention not only significantly reduces the cost of vehicle torque vectoring tuning, but also, under conditions such as low slip angles, exhibits more linear and smooth performance at the vehicle's dynamic limits, and is more efficient than differential torque vectoring.

[0041] Compared to the grip-type torque vectoring model, the present invention is based on a more comprehensive simplified vehicle dynamics model rather than an assumption about vehicle grip or a specific operating condition itself. Therefore, the present invention performs better under more and wider operating conditions and will not cause the vehicle to quickly lose controllability due to vehicle deviation from a specific operating condition. In addition, since the present invention can greatly improve the operating window and capability range of ABS / TC, compared to the grip-type torque vectoring model, the present invention can also greatly expand the fault tolerance and dynamic robustness of the vehicle's dynamic limits by applying corresponding torque vectoring.

[0042] Compared with the AI large-model torque vectoring model, the present invention not only tends to be conservative and stable in longitudinal torque distribution, but also has multiple built-in protections. The driver's subjective will always takes precedence over the torque vectoring algorithm, and its safety and controllability are also greatly improved.

[0043] Compared with all existing torque vectoring models, the present invention has extremely high adaptability to vehicles with different mechanical structures, different hardware modes, different dynamic characteristics, and different external conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 A schematic flow chart of a torque vector prediction method based on a simplified vehicle dynamics model of the present invention;

[0045] Figure 2 Schematic diagram of the process of obtaining the driver prediction vector in an embodiment of the present invention;

[0046] Figure 3 Schematic diagram of the principle of difference operation in an embodiment of the present invention. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention is further described below with reference to the accompanying drawings.

[0048] In this embodiment, as shown in Figure 1 , the present invention proposes a torque vector prediction method based on a simplified vehicle dynamics model, including the steps of:

[0049] S10. Obtain the real-time body attitude and wheel speed data of the vehicle in the vehicle bus; use a composite sensor to obtain the real-time motion state and direction of the vehicle, as well as the real-time external acceleration; based on the vehicle dynamics model, obtain a simplified vehicle dynamics model and calculate the existing path of the vehicle;

[0050] S20. Obtain the steering input data and pedal input data, and use the driving input prediction model to obtain the driver prediction vector;

[0051] S30. After performing a difference operation on the calculated existing path of the vehicle and the driver prediction vector, use the torque vector calculation model to obtain the torque vector output result.

[0052] As an optimized solution of the above embodiment, in the simplified vehicle dynamics model:

[0053] By combining the suspension motion data obtained by the suspension sensor, the real-time geometric position and deformation amount of the inner wall of the wheel, the wheel size, the geometric position of the vehicle center of gravity including two parts: the center of gravity height CGH and the vehicle corner weight, the vehicle acceleration, the spring value of the suspension, the data collected by the four-wheel wheel speed sensors of the vehicle obtained from the bus, the driver's steering and pedal inputs, and the equivalent grip benefit value corresponding table of two spring-damping models and the corresponding tire deformation conditions to form a simplified tire model, the actual motion trajectory of the vehicle, the length, width, and height of the vehicle, as well as the wheelbase and track width, to construct the simplified vehicle dynamics model during the vehicle motion process.

[0054] The construction process of the simplified vehicle dynamics model includes the steps of:

[0055] S101. According to the suspension motion data obtained by the sensor during the vehicle motion, the geometric position of the inner wall of the wheel, and the wheel size, and according to the geometric appearance of the suspension, calculate the wheel ground contact midpoint position P_CP, and obtain the velocity instantaneous center P_IC of any motion state of the suspension and the suspension rotation center P_RC;

[0056] S102. By comprehensively obtaining the suspension rotation center P_RC, obtain the roll rotation axis of the vehicle at any time; according to the roll rotation axis and the real-time acceleration of the vehicle, calculate the original angular momentum of the vehicle at any time;

[0057] S103. Substitute the actual transverse and longitudinal moments of inertia into the suspension motion data according to the suspension motion data; since the spring value of the suspension and the geometric position of the suspension system are known, obtain the effective damping value in any motion state of the suspension and the suspension speed thresholds for the fast and slow valve switches.

[0058] S104. Construct a vehicle suspension model: perform a real-time force analysis on the vehicle suspension geometric structure based on the real-time yaw and pitch data of the vehicle itself, so as to obtain the internal stress of the suspension system. Through the internal stress of the suspension and the tracking of the deformation amount of the suspension bushing achieved according to visual tracking, perform an independent force analysis on each bushing, and then calculate the equivalent hardness value of the suspension flexible component according to its deformation amount under different external forces.

[0059] S106. Take the first derivative of the constructed vehicle suspension model. Through the force analysis obtained by combining the suspension geometric structure, obtain the real-time tire load, substitute the accelerations on the x, y, and z axes of the vehicle and the vehicle sideslip angle obtained from the vehicle motion state, decompose the overall vehicle grip vector, so as to obtain the real-time grip estimation on each wheel, and calculate the tire slip ratio κ according to the difference between the wheel speed sensor and the calculated ground speed.

[0060] S106. Take the first derivative of the constructed vehicle suspension model. Through the force analysis obtained by combining the suspension geometric structure, obtain the real-time tire load, substitute the accelerations on the x, y, and z axes of the vehicle and the vehicle sideslip angle obtained from the vehicle motion state, decompose the overall vehicle grip vector, so as to obtain the real-time grip estimation on each wheel, and calculate the tire slip ratio κ according to the difference between the wheel speed sensor and the calculated ground speed.

[0061] S107. According to the total load W Transfer,sum acting on the tire and the four-wheel lateral grip F y , deduce a grip equivalent μ corresponding to the wheel deformation amount, where

[0062] μ = F y / W Transfer,sum ·κ;

[0063] Form the first part of the simplified tire model, that is, the tire horizontal grip empirical model.

[0064] S108. Since the suspension system load and the wheel deformation amount are known, regard the whole wheel as two longitudinal spring-damper structures connected by the tread, and calculate the equivalent tire load-lateral grip corresponding relationship according to the external force exerted by the suspension on the tire and the actual deformation amount of the tire itself, forming the second part of the simplified tire model, that is, the tire force empirical model.

[0065] S109. Based on the suspension motion data measured in S101, the original angular momentum obtained in S102, the suspension speed threshold for the fast and slow valve opening and closing obtained in S103, the equivalent hardness value obtained in S104, the tire horizontal grip empirical model obtained in S107, and the tire force empirical model obtained in S108, a vehicle dynamics model is comprehensively obtained.

[0066] From the simplified vehicle dynamics model, the existing path of the vehicle is deduced, including:

[0067] Assume that the vehicle does not use any active power distribution to assist steering and only relies on the lateral grip of the vehicle under the current conditions for steering. Its motion trajectory is obtained by relying on the simplified vehicle dynamics model to perform mechanical calculations for the next moment to obtain the existing path of the vehicle.

[0068] As an optimized solution to the above embodiment, as Figure 2 shown, in the driving input prediction model:

[0069] Through the offset between the absolute value of the steering wheel angle and the current pointing state of the vehicle, the acceleration of the steering wheel angle, the output value corresponding to the preset throttle curve of the absolute value of the pedal input, the input acceleration of the pedal, and the preset longitudinal torque vector offset comparison table, a vector value of the vehicle running direction expected by the driver himself is comprehensively deduced, that is, the driver prediction vector output

[0070] Preferably, in the driving input prediction model:

[0071] When the vehicle discovers through GPS that there is an upcoming sharp turn that requires heavy braking ahead, it will choose to reduce the offset weight of the driver prediction vector D and increase the weight in the acceleration and deceleration directions of D during the rapid deceleration process, so as to reduce the magnitude of the torque vector difference TV and increase the handling stability characteristics of the vehicle during the deceleration process; similarly, if the system discovers that the vehicle is approaching a corner, it will increase the offset weight of D and reduce its weight in the acceleration and deceleration directions, so as to reduce the handling stability characteristics under specific conditions in exchange for better turning characteristics.

[0072] The present invention can comprehensively judge the state of the vehicle, such as cornering, rapid acceleration / rapid deceleration, normal driving, emergency lane change, etc., according to the map or pre-programmed form, and the road condition information ahead provided by other sensors including the vehicle itself (such as lidar, intelligent driving camera, etc.), and perform a certain degree of offset and correction on the driver prediction vector value, so as to make a more comprehensive judgment according to the preset scenario mode.

[0073] [[ID=,26]]Preferably, the system compares the un-offset D vector with the offset DP vector. If the offset amount between the two is higher than a certain preset value, the system will calculate the torque vector based on the D vector.

[0074] The present invention also compares the driver prediction vector value without the offset in part Ⅳ. If the difference between the two exceeds a preset threshold, the system will use the driver prediction vector value without offset as the basis for actual output.

[0075] As an optimized solution of the above embodiment, after calculating the difference between the inferred existing path of the vehicle and the driver prediction vector (as shown in Figure 3 ), the torque vector output result is obtained by using the torque vector calculation model.

[0076] In the torque vector calculation model:

[0077] Ⅰ: The real-time pointing, speed and acceleration of the whole vehicle, and the differences in pointing, speed and acceleration between it and the driver prediction vector obtained by the driving input prediction model of model ②; Ⅱ: The real-time available grip of the four wheels output by the simplified vehicle dynamics model; Ⅲ: The real-time attitude and rudder angle information of the vehicle;

[0078] Using the data obtained in Ⅰ, Ⅱ and Ⅲ, according to the calculated real-time available grip of the tires, the differences obtained in Ⅰ are respectively tried to be solved on the four wheels, so as to obtain the final specific output mode of the torque vector.

[0079] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of the present invention claimed is defined by the appended claims and their equivalents.

Claims

1. A torque vector prediction method based on a simplified vehicle dynamics model, characterized in that, Including the steps: S10, obtaining the real-time body attitude and wheel speed data of the vehicle in the vehicle bus; using a composite sensor to obtain the real-time motion state and direction of the vehicle, as well as the real-time external acceleration; Based on the vehicle dynamics model, obtaining a simplified vehicle dynamics model and calculating the existing path of the vehicle; S20, obtaining the steering input data and pedal input data, and using a driving input prediction model to obtain the driver prediction vector; S30, after performing a difference operation between the calculated existing path of the vehicle and the driver prediction vector, using a torque vector calculation model to obtain the torque vector output result.

2. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 1, characterized in that, In the simplified vehicle dynamics model: By combining the suspension motion data obtained by the suspension sensor, the real-time geometric position and deformation amount of the inner wall of the wheel, the wheel size, the geometric position of the vehicle center of gravity including two parts: the center of gravity height CGH and the vehicle corner weight, the vehicle acceleration, the spring value of the suspension, the data collected by the four-wheel wheel speed sensors of the vehicle obtained from the bus, the driver's steering and pedal inputs, two spring-damping models and the equivalent grip benefit value corresponding table of the corresponding tire deformation conditions to form a simplified tire model, the actual motion trajectory of the vehicle, the length, width, height, wheelbase and track width of the vehicle, to construct the simplified vehicle dynamics model during the vehicle motion process.

3. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 2, characterized in that, The process of constructing the simplified vehicle dynamics model includes the steps: S101, according to the suspension motion data, the geometric position of the inner wall of the wheel and the wheel size obtained by the sensor during the vehicle motion, and according to the geometric appearance of the suspension, calculating the wheel ground contact midpoint position P_CP, obtaining the instantaneous velocity center P_IC of any motion state of the suspension and the suspension rotation center P_RC; S102, by synthesizing the suspension rotation center P_RC, obtaining the roll rotation axis of the vehicle at any time; according to the roll rotation axis and the real-time acceleration of the vehicle, calculating the original angular momentum of the vehicle at any time; S103, according to the suspension motion data, substituting the actual transverse and longitudinal moments of inertia into the suspension motion data; since the spring value of the suspension and the geometric position of the suspension system are known, obtaining the effective damping value of any motion condition of the suspension and the suspension speed threshold of the fast and slow valve switches; S104, constructing a vehicle suspension model: performing a real-time force analysis on the geometric structure of the vehicle suspension according to the real-time yaw and pitch data of the vehicle itself, thereby obtaining the internal stress of the suspension system, and through the internal stress of the suspension and the tracking of the suspension bushing deformation amount realized by visual tracking, performing an independent force analysis on each bushing, and then calculating the equivalent hardness value of the suspension flexible member according to its deformation amount under different external forces; S105, using the measured suspension motion data, the original angular momentum, the suspension speed threshold of the fast and slow valve switches and the equivalent hardness value, obtaining all the forces acting on the tire at any time, including the forces caused by the vehicle center of gravity transfer, the forces caused by the mutual traction of the rigid components inside the suspension system, and the forces caused by the deformation of the suspension flexible member; S106, perform a first derivative on the constructed vehicle suspension model, and obtain the real-time tire load W through the force analysis combined with the suspension geometric structure. transfer , substitute the accelerations on the X, Y, and Z axes of the vehicle and the vehicle sideslip angle obtained from the vehicle motion state, decompose the overall vehicle grip vector, so as to obtain the real-time grip estimation on each wheel, and calculate the tire slip ratio κ according to the difference between the wheel speed sensor and the calculated ground speed. S107. Based on the total load W on the scope tires Transfer,sum and the four-wheel lateral grip F y a grip equivalent μ corresponding to the wheel deformation is derived according to the corresponding relationship, where Forming the first part of the simplified tire model, that is, the tire horizontal grip empirical model; S108. Since the load of the suspension system and the wheel deformation amount are known, the wheel as a whole is regarded as two longitudinal spring-damper structures connected by the tread. According to the external force exerted by the suspension on the tire and the actual deformation amount of the tire itself, the corresponding relationship between the equivalent tire deformation amount and the tire force is calculated to form the second part of the simplified tire model, that is, the tire force empirical model. S109. Based on the suspension motion data measured in S101, the original angular momentum obtained in S102, the suspension speed threshold of the fast and slow valve switches obtained in S103, the equivalent hardness value obtained in S104, the tire horizontal grip empirical model obtained in S107, and the tire force empirical model obtained in S108, a vehicle dynamics model is comprehensively obtained.

4. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 1, characterized in that, From the simplified vehicle dynamics model, the existing path of the vehicle is deduced, including: Assume that the vehicle does not use any active power distribution to assist steering and only relies on the lateral grip of the vehicle under the current conditions to steer. Its motion trajectory is obtained by performing mechanical calculations for the next moment based on the simplified vehicle dynamics model to obtain the existing path of the vehicle.

5. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 1, characterized in that, In the driving input prediction model: Through the offset between the absolute value of the steering wheel angle and the current pointing state of the vehicle, the acceleration of the steering wheel angle, the output value of the pedal input absolute value corresponding to the preset throttle curve, the input acceleration of the pedal, and the preset longitudinal torque vector offset comparison table, a vector value of the vehicle running direction expected by the driver himself is comprehensively deduced, that is, the driver prediction vector output.

6. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 3, characterized in that, In the driving input prediction model: When the vehicle discovers through GPS that there will be a sharp turn ahead that requires heavy braking, it will choose to reduce the offset weight of the driver prediction vector D and increase the weight in the acceleration and deceleration directions of D during the rapid deceleration process, so as to reduce the magnitude of the torque vector difference TV and increase the handling and stability characteristics of the vehicle during the deceleration process. Similarly, if the system discovers that the vehicle is about to approach a corner, it will increase the offset weight of D and reduce its weight in the acceleration and deceleration directions, so as to reduce the handling and stability characteristics under specific conditions in exchange for better turning characteristics.

7. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 6, characterized in that, The system compares the un-offset D vector with the offset DP vector. If the offset amount between the two is higher than a certain preset value, the system will calculate the torque vector based on the D vector.

8. A torque vector prediction method based on a simplified vehicle dynamics model according to claim 1, characterized in that, In the torque vector calculation model: Ⅰ: The real-time pointing, speed, and acceleration of the vehicle as a whole, and the differences in pointing, speed, and acceleration between it and the driver prediction vector obtained from the driving input prediction model of the ② model; Ⅱ: The real-time available grip of the four wheels output by the simplified vehicle dynamics model; Ⅲ: The real-time attitude and rudder angle information of the vehicle; Using the data obtained in Ⅰ, Ⅱ, and Ⅲ, the available grip of the tire obtained according to the calculation is respectively used to solve the differences obtained in Ⅰ on the four wheels, so as to obtain the final specific output mode of the torque vector.

Images