Data-driven high-precision and high-real-time suspension dynamic system and construction method

Abstract

The invention discloses a suspension dynamics system and a construction method, and the method comprises the steps: carrying out the recursion of a multi-body model and a neural network model, carrying out the equivalent transformation through introducing an auxiliary rigid body and a corresponding kinematic pair, and simplifying the kinematic pair in an actual vehicle multi-body system; cutting a preset position of the vehicle multi-body system to convert a closed-loop topological structure of the vehicle multi-body system into an open-loop structure comprising a plurality of branch chains, and adding a constraint equation to realize the equivalence of the open-loop structure and the original closed-loop topological structure; the recursive multi-body model is a multi-body dynamic system described in an ordinary differential equation form, and the neural network model is used for obtaining C characteristic parameters of the suspension according to tire force. According to the method, the dynamic model of the vehicle multi-body system and the K / C multi-direction coupling characteristics of the suspension system are integrated, the data-driven vehicle multi-body system model, namely the suspension dynamic system, is obtained, and the calculation efficiency and the solving precision can be effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of vehicle suspension dynamics, and in particular to a data-driven high-precision and high-real-time suspension dynamics system and a construction method thereof. Background Art

[0002] With the continuous advancement of computer and software technology, virtual prototype dynamic simulation technology, centered on multibody system dynamics, has been gradually applied to vehicle design, shortening vehicle R&D cycles, reducing development costs, and improving product design and manufacturing quality. The suspension system is the collective term for all force-transmitting connections between a vehicle's frame and axles or wheels. Its function is to transmit forces and torques acting between the wheels and frame, cushion the impact forces transmitted to the frame or body by uneven road surfaces, and attenuate the resulting vibrations to ensure smooth vehicle operation. Existing simulation analyses often describe the vehicle's multibody dynamics system using differential algebraic equations, which are complex to solve and lack real-time performance. Furthermore, the suspension's K&C characteristics (Kinematic and Compliance) do not fully satisfy the linear superposition assumption under multi-directional coupling conditions. This results in models that do not accurately fit the vehicle's actual conditions, especially under multi-directional coupling conditions. Summary of the Invention

[0003] In response to the above-mentioned defects or improvement needs of the prior art, the present invention provides a suspension dynamics system and construction method, which is a data-driven, high-precision, and high-real-time suspension dynamics system and construction method, which can obtain sufficient calculation speed and solution accuracy.

[0004] To achieve the above objectives, the present invention adopts the following technical solutions.

[0005] In one aspect, a suspension dynamics system is provided, comprising:

[0006] A recursive multi-body model, which simplifies the kinematic pairs in an actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation; and by shearing the vehicle multi-body system at preset positions, transforms the closed-loop topology of the vehicle multi-body system into an open-loop structure containing multiple branch chains, and achieves equivalence between the open-loop structure and the original closed-loop topology by adding constraint equations;

[0007] The recursive multi-body model is a multi-body dynamics system described in the form of ordinary differential equations. The multi-body dynamics system described in the form of ordinary differential equations is obtained by the following method:

[0008] Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates;

[0009] The dependent hinge coordinates are expressed as independent hinge coordinates, and the dynamic system is described by the vehicle body quasi-velocity and the independent hinge coordinates;

[0010] The multi-body dynamics system described in the form of differential algebraic equations is transformed into a multi-body dynamics system described in the form of ordinary differential equations;

[0011] The recursive multi-body model is used to obtain the K characteristic parameters of the suspension according to the tire force;

[0012] The suspension dynamics system further includes: a neural network model, the neural network model being used to obtain C characteristic parameters of the suspension based on tire forces;

[0013] The recursive multi-body model is also used to obtain the vehicle's posture change parameters based on the K characteristic parameters and the C characteristic parameters.

[0014] In some embodiments, the suspension dynamics system further includes a tire model, wherein the tire model is used to obtain tire force according to tire alignment parameters, wherein the tire alignment parameters are obtained by summing the K characteristic parameter and the C characteristic parameter.

[0015] In some embodiments, the recursive multi-body model simplifies the kinematic pairs in the actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, including:

[0016] The bushing connecting the upper end of the shock absorber to the vehicle frame is simplified to a three-degree-of-freedom ball joint, ignoring the degree of freedom in one direction. A first auxiliary rigid body is introduced, with two mutually perpendicular revolute pairs at both ends, thereby replacing the ball joint with two single-degree-of-freedom revolute pairs.

[0017] There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body is introduced, thereby replacing the cylindrical pair with a rotational pair and a translation pair.

[0018] In some embodiments, the neural network model includes a fully connected neural network or a recurrent neural network;

[0019] For the fully connected neural network, the input is a column vector consisting of the longitudinal force, lateral force, and self-aligning moment of the tire at the current moment and several previous moments, and the output is the tire positioning parameter under the coupled working condition;

[0020] For the recurrent neural network, the input is a column vector consisting of the longitudinal force, lateral force, and self-aligning torque of the tire at the current moment, and the output is the tire positioning parameters under the coupled working condition.

[0021] In some embodiments, the suspension dynamics system uses the Adams-Bashforth-Moulton algorithm as a benchmark algorithm to improve the real-time performance of simulation calculations.

[0022] On the other hand, a suspension dynamics system construction method is also provided, the suspension dynamics system construction method comprising: establishing a recursive multi-body model and a neural network model;

[0023] Building a recursive multi-body model involves:

[0024] By introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, the kinematic pairs in the actual vehicle multi-body system are simplified; and by shearing the preset positions of the vehicle multi-body system, the closed-loop topology structure of the vehicle multi-body system is transformed into an open-loop structure containing multiple branch chains, and the equivalence of the open-loop structure to the original closed-loop topology structure is achieved by adding constraint equations;

[0025] Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates;

[0026] The dependent hinge coordinates are expressed as independent hinge coordinates, and the dynamic system is described by the vehicle body quasi-velocity and the independent hinge coordinates;

[0027] Transform the multi-body dynamics system described in the form of differential algebraic equations into a multi-body dynamics system described in the form of ordinary differential equations;

[0028] Wherein, the neural network model includes a fully connected neural network or a recurrent neural network.

[0029] In some embodiments, establishing the fully connected neural network includes: performing hysteresis removal processing on data containing hysteresis curves, using the hysteresis-removed data for training, adjusting network parameters to prevent overfitting, and selecting the network structure and network parameters with the best fitting effect;

[0030] Establishing the recurrent neural network includes: retaining curve hysteresis information in experimental or simulated data, optimizing the network structure and scale, and obtaining a preset simulation accuracy at a minimum scale.

[0031] In some embodiments, the suspension dynamics system construction method further includes establishing a tire model, wherein the tire model adopts a magic formula tire model, and the tire model is used to obtain tire force according to tire positioning parameters.

[0032] In some embodiments, the kinematic pairs in the actual vehicle multi-body system are simplified by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, including:

[0033] The bushing connecting the upper end of the shock absorber to the vehicle frame is simplified to a three-degree-of-freedom ball joint, ignoring the degree of freedom in one direction. A first auxiliary rigid body is introduced, with two mutually perpendicular revolute pairs at both ends, thereby replacing the ball joint with two single-degree-of-freedom revolute pairs.

[0034] There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body is introduced, thereby replacing the cylindrical pair with a rotational pair and a translation pair.

[0035] The present application also provides a method for dynamic simulation of a suspension system, wherein the method performs simulation using the suspension dynamic system as described in any one of the above items.

[0036] Through the method proposed in the present invention, a vehicle multibody system dynamics model characterized by a mixture of quasi-velocity and relative hinge coordinates is established based on a two-step recursive method, and the DAE is converted into an ODE for solution. Furthermore, by using two neural networks, a fully connected neural network and a recurrent neural network trained with data obtained from multidirectional coupling experiments and simulations, the fitting relationship between the K / C characteristics of the suspension under multidirectional coupling conditions and the tire alignment parameters is obtained. The present invention integrates the vehicle multibody system dynamics model with the K / C multidirectional coupling characteristics of the suspension system to obtain a data-driven vehicle multibody system model, namely the suspension dynamics system. The use of the suspension dynamics system proposed in the present invention can effectively improve computational efficiency and solution accuracy. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 It is a schematic structural diagram of a suspension dynamics system according to an embodiment of the present invention.

[0038] Figure 2 It is a schematic diagram of a suspension dynamics system and a flow chart of an embodiment of the present invention.

[0039] Figure 3 Schematic diagram of a recursive multi-body model and its construction method according to an embodiment of the present invention.

[0040] Figure 4 Schematic diagram of a fully connected neural network according to an embodiment of the present invention.

[0041] Figure 5 Schematic diagram of a recurrent neural network according to an embodiment of the present invention.

[0042] Figure 6It is an overall schematic diagram of a suspension dynamics system according to an embodiment of the present invention.

[0043] Figure 7 It is a schematic diagram of the relationship between the vehicle body coordinate system and the absolute coordinate system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0045] In the description of the present invention, it should be understood that the terms "upper", "lower", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "circumferential", etc., indicating orientations or positional relationships, are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The interpretation of such words should be made from the perspective of those skilled in the art. For example, "above..." and "below..." should be understood as the positional relationship of the main structure or structure of the component, etc. in the initial state, which may break through the above positional relationship during movement. "...disposed on..." should be understood as the approximate connection relationship of the components, rather than necessarily being above.

[0046] In the present invention, unless otherwise expressly defined or limited, the terms "disposed," "installed," "connected," "connected," "fixed," etc. should be understood broadly from the perspective of those skilled in the art. For example, they may refer to fixed connections, detachable connections, or integration; they may refer to direct connections or indirect connections through an intermediate medium; they may refer to internal communication between two elements or interaction between two elements, unless otherwise expressly defined. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.

[0047] In the description of this specification, the reference terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" mean that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0048] The suspension's K characteristic, or kinematic characteristic, describes the variations in tire alignment parameters caused by suspension motion and is primarily related to the geometric parameters of the suspension's guiding mechanism. For vehicle suspension, the most fundamental motion is wheel hop (in-direction wheel hop) when the vehicle encounters uneven road surfaces during forward motion, or roll (in-direction wheel hop) when the vehicle turns. In short, the suspension's K characteristic is the kinematic behavior exhibited by wheel hop.

[0049] The suspension's C-characteristic, or Compliance elastic kinematic characteristic, describes the variations in tire alignment parameters caused by the forces and moments between the tire and the road. It is primarily related to the suspension's elastic elements and elastic properties. Rubber bushings deform when subjected to external forces, and the stiffness matching between different bushings affects the suspension's characteristics, and thus the vehicle's handling stability. Unlike the K-characteristic, the C-characteristic describes the changes in suspension parameters when forces and moments are applied to the tire under a fixed wheel hop (corresponding to a certain suspension load condition).

[0050] The K and C characteristics interact with each other, forming the basis of suspension performance. Conventional K / C testing typically involves applying a fixed wheel hop to the left and right sides of a test bench, and measuring the changes in the suspension's K and C characteristics using experimental equipment.

[0051] The present invention provides a suspension dynamics system and construction method, which is a data-driven, high-precision, and high-real-time suspension dynamics system and construction method, which can obtain a suspension dynamics system with sufficient calculation speed and solution accuracy.

[0052] Figure 1 FIG1 is a schematic diagram of the structure of a suspension dynamics system according to an embodiment of the present invention. Figure 1 , the suspension dynamics system includes:

[0053] A recursive multi-body model, which simplifies the kinematic pairs in an actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation; and by shearing the vehicle multi-body system at preset positions, transforms the closed-loop topology of the vehicle multi-body system into an open-loop structure containing multiple branch chains, and achieves equivalence between the open-loop structure and the original closed-loop topology by adding constraint equations;

[0054] The recursive multi-body model is a multi-body dynamics system described in the form of ordinary differential equations. The multi-body dynamics system described in the form of ordinary differential equations is obtained by the following method:

[0055] Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates;

[0056] The dependent hinge coordinates are expressed as independent hinge coordinates, and the dynamic system is described by the vehicle body quasi-velocity and the independent hinge coordinates;

[0057] The multi-body dynamics system described in the form of differential algebraic equations is transformed into a multi-body dynamics system described in the form of ordinary differential equations;

[0058] The recursive multi-body model is used to obtain the K characteristic parameters of the suspension according to the tire force;

[0059] The suspension dynamics system further includes: a neural network model, the neural network model being used to obtain C characteristic parameters of the suspension based on tire forces;

[0060] The recursive multi-body model is also used to obtain the vehicle's posture change parameters based on the K characteristic parameters and the C characteristic parameters.

[0061] In the embodiment of the present application, the posture change of the entire vehicle is finally obtained through multiple cycles of calculation. In the embodiment of the present application, a multi-body dynamics modeling method based on the quasi-velocity and relative hinge coordinate expression of the two-step recursive method is established to establish a recursive multi-body model, eliminate the constraint equations, and convert the DAE (differential algebraic equation) into the ODE (ordinary differential equation), thereby improving the solution efficiency. At the same time, the K&C characteristics of the suspension are integrated into the suspension dynamics system through a data-driven high-precision characterization algorithm, namely a neural network model, which further improves the simulation accuracy of the real-time recursive multi-body model.

[0062] Figure 2 This is a schematic diagram of the suspension dynamics system and flow chart of an embodiment of the present invention. Figure 2In some embodiments, the suspension dynamics system further includes a tire model, which is used to obtain tire forces based on tire alignment parameters (also known as wheel alignment parameters), wherein the tire alignment parameters are obtained by summing the K characteristic parameters and the C characteristic parameters. Tire alignment parameters primarily include lateral displacement, longitudinal displacement, caster angle, camber angle, toe angle, etc. The tire model can calculate the longitudinal slip rate and sideslip angle based on the tire alignment parameters combined with the relative position and posture of the wheel and the road surface. The tire forces can then be calculated using the longitudinal slip rate and sideslip angle.

[0063] The K characteristic is the change in tire alignment parameters caused by the vertical movement of the suspension. The tire alignment parameters include lateral displacement, longitudinal displacement, caster angle, camber angle, and toe angle, from which the K characteristic parameters can be obtained; the C characteristic is the change in tire alignment parameters caused by tire force. The tire alignment parameters also include lateral displacement, longitudinal displacement, caster angle, camber angle, and toe angle. The tire force specifically includes longitudinal force, lateral force, and self-aligning torque, from which the C characteristic parameters can be obtained. Suspension bounce (i.e., vertical movement of the suspension) and tire force will both cause changes in tire alignment parameters. When applied, each specific wheel angle is affected by both suspension bounce and tire force. These two effects are added together to obtain the actual wheel angle change.

[0064] exist Figure 2 In the illustrated embodiment, the tire model is a tire model based on the Magic Formula, namely, the MF model.

[0065] Figure 3 Schematic diagram of a recursive multi-body model and its construction method according to an embodiment of the present invention. Figure 3 Taking the McPherson suspension as an example, it includes a sleeve at the upper end of the shock absorber ( Figure 3 ), steering rack, steering knuckle, steering tie rod, control arm, bushing (not shown) used to connect the upper end of the shock absorber and the frame (not shown). Steering knuckle is connected to the wheel ( Figure 3 (For example, the right front wheel is used in this example), the wheel can be considered part of the suspension. The steering rod is a lightweight rod. The frame is fixedly connected to the vehicle body. In some embodiments, the recursive multi-body model simplifies the kinematic pairs in the actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, including:

[0066] The bushing used to connect the upper end of the shock absorber and the frame is simplified to a three-degree-of-freedom ball pair, ignoring the degree of freedom in one direction, and introducing the first auxiliary rigid body ( Figure 3 The auxiliary rigid body 1 shown in FIG, wherein the two ends of the first auxiliary rigid body are mutually perpendicular rotation pairs, thereby replacing the spherical pair with two single-degree-of-freedom rotation pairs;

[0067] There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body ( Figure 3 The auxiliary rigid body 2) is shown, thereby replacing the cylindrical pair with a rotation pair and a translation pair.

[0068] Continue to refer Figure 3 , by cutting the preset position of the vehicle multi-body system, the closed-loop topology structure of the vehicle multi-body system is converted into an open-loop structure containing multiple branch chains, and the open-loop structure is made equivalent to the original closed-loop topology structure by adding constraint equations, including:

[0069] The ball joint between the control arm and the steering knuckle is cut, and the lightweight steering tie rod between the steering rack and the steering knuckle and the ball joints at both ends of the steering tie rod are removed. The closed-loop topology structure of the vehicle multi-body system is opened, and the closed-loop McPherson suspension system is transformed into an open-loop structure with three branch chains. Figure 3 As shown, the three branches are: the first branch consisting of the vehicle body - steering rack, the second branch consisting of the vehicle body - control arm, and the third branch consisting of the vehicle body - auxiliary rigid body 1 - auxiliary rigid body 2 - sleeve - steering knuckle - right front wheel. The original closed-loop structure is then made equivalent to the open-loop structure by adding constraint equations. The added constraint equations correspond to the ball joint constraints between the control arm and steering knuckle, and the ball joint constraints at both ends of the steering rod.

[0070] In some embodiments, the neural network model includes a fully connected neural network or a recurrent neural network.

[0071] Figure 4 Schematic diagram of a fully connected neural network according to an embodiment of the present invention. Figure 4 The fully connected neural network takes as input a column vector consisting of the tire's longitudinal force, lateral force, and aligning torque at the current moment and several previous moments, and outputs the tire alignment parameters under the coupled working condition. These tire alignment parameters include toe angle change, wheel camber, wheel caster, wheel center longitudinal displacement, and wheel center lateral displacement.

[0072] Figure 5 Schematic diagram of a recurrent neural network according to an embodiment of the present invention. Figure 5 The recurrent neural network takes as input a column vector consisting of the tire's longitudinal force, lateral force, and aligning torque at the current moment, and outputs the tire alignment parameters under the coupled working condition. These tire alignment parameters include toe angle change, wheel camber, wheel caster, wheel center longitudinal displacement, and wheel center lateral displacement.

[0073] In some embodiments, the suspension dynamics system uses the Adams-Bashforth-Moulton algorithm as a benchmark algorithm to improve the real-time performance of the simulation calculation. The Adams-Bashforth-Moulton algorithm is used for recursive multi-body model simulation and has good simulation results.

[0074] Figure 6 It is an overall schematic diagram of a suspension dynamics system according to an embodiment of the present invention. Figure 7 It is a schematic diagram of the relationship between the vehicle body coordinate system and the absolute coordinate system in one embodiment of the present invention.

[0075] refer to Figure 7 In some embodiments of the present invention, the quasi-speed relationship corresponding to the vehicle body quasi-speed includes the relationship between the vehicle body quasi-speed describing the vehicle body movement and the vehicle body absolute speed, and the relationship between the quasi-angular velocity describing the vehicle body rotational motion and the vehicle body absolute angular velocity.

[0076] The origin of the body coordinate system is set at the center of mass of the body, the direction perpendicular to the lower plane of the frame pointing to the roof is the Z' axis direction, the direction from the center of mass to the rear of the body is the X' direction, and following the right-hand principle, the direction pointing to the right side of the body is the Y' direction, resulting in the body coordinate system O-X'Y'Z'.

[0077] The transformation relationship between the vehicle body coordinate system and the absolute coordinate system is expressed by the transformation matrix in the form of Euler angles. The absolute coordinate system is Figure 7 The global coordinate system O-XYZ is shown in .

[0078] Let pth, yaw, and rll be the Euler angles of the vehicle body coordinate system relative to the global coordinate system, which are the pitch angle, yaw angle, and roll angle of the vehicle body respectively. The rotation matrix A is expressed as follows:

[0079]

[0080] Project the absolute velocity of the vehicle body into the vehicle body coordinate system. The projections in the X', Y', and Z' directions are u, v, and w respectively. The conversion relationship is as follows:

[0081]

[0082] where x 1 、y 1 、z 1 are the displacements of the vehicle body in the global coordinate system, A is the rotation matrix between the vehicle body coordinate system and the global coordinate system, and t is the time.

[0083] Project the absolute angular velocity of the vehicle body into the vehicle body coordinate system. The projections in the X', Y', and Z' directions are p, q, and r respectively. The conversion relationship is as follows:

[0084]

[0085] Among them, the transformation matrix D -1 for:

[0086]

[0087] The present invention uses quasi-velocity to represent the dynamic equations of a closed-loop multi-body system, which facilitates the description of the motion of the system and further transformation, while also making it easier to define and obtain the forces acting on the vehicle body.

[0088] In an embodiment of the present application, through two speed transformations, the multi-body dynamics system described in the form of differential algebraic equations (DAE) is converted into a multi-body dynamics system described in the form of ordinary differential equations (ODE), that is, the recursive multi-body model is a multi-body dynamics system described in the form of ordinary differential equations (ODE).

[0089] Specifically, the degrees of freedom of each component in a closed-loop multibody system can be defined based on quasi-velocities. To reduce the degrees of freedom of the dynamic model, the multibody dynamics system described entirely by quasi-velocities is transformed into a dynamics system described jointly by the quasi-velocities of the vehicle body and the hinge coordinates. This is the first velocity transformation. The specific transformation method is as follows:

[0090]

[0091] in is the quasi-velocity of all components, is the derivative of the hinge coordinates, and R is the velocity conversion matrix. In fact, the hinge coordinates of the closed-loop multi-body system are not completely independent. The hinge coordinates can be divided into independent coordinates and non-independent coordinates. d With independent hinge coordinate z i Indicates that the multi-body dynamics system is described by the body quasi-velocity and independent hinge coordinates. This is the second velocity transformation. The specific transformation method is as follows:

[0092]

[0093] in, is the dependent coordinate part of the constraint Jacobian matrix, is the independent coordinate part of the constraint Jacobian matrix.

[0094] By the above two velocity transformations, the constraint equations of the multi-body dynamics system are eliminated, and the multi-body dynamics system originally described by differential algebraic equations is transformed into a multi-body dynamics system described by ordinary differential equations. The equations are as follows:

[0095]

[0096] Among them, M, Qex , Q in They are the transformed mass matrix, external force and velocity coupling force respectively.

[0097] In the embodiment of the present invention, a multi-body system in the form of DAE is converted into a multi-body system in the form of ODE through two velocity transformations, thereby avoiding problems that may be caused by constraint violation, while also reducing the degree of freedom of the system and improving solution efficiency and calculation accuracy.

[0098] Some embodiments of the present application further provide a suspension dynamics system construction method, which is used to establish the suspension dynamics system in any of the above embodiments.

[0099] The suspension dynamics system construction method includes: establishing a recursive multi-body model and a neural network model;

[0100] Building a recursive multi-body model involves:

[0101] By introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, the kinematic pairs in the actual vehicle multi-body system are simplified; and by shearing the preset positions of the vehicle multi-body system, the closed-loop topology structure of the vehicle multi-body system is transformed into an open-loop structure containing multiple branch chains, and the equivalence of the open-loop structure to the original closed-loop topology structure is achieved by adding constraint equations;

[0102] Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates;

[0103] The dependent hinge coordinates are expressed as independent hinge coordinates, and the dynamic system is described by the vehicle body quasi-velocity and the independent hinge coordinates;

[0104] Transform the multi-body dynamics system described in the form of differential algebraic equations into a multi-body dynamics system described in the form of ordinary differential equations;

[0105] Wherein, the neural network model includes a fully connected neural network or a recurrent neural network.

[0106] In some embodiments, establishing the fully connected neural network includes: performing hysteresis removal processing on data containing hysteresis curves, using the hysteresis-removed data for training, adjusting network parameters to prevent overfitting, and selecting the network structure and network parameters with the best fitting effect;

[0107] Establishing the recurrent neural network includes: retaining curve hysteresis information in experimental or simulated data, optimizing the network structure and scale, and obtaining a preset simulation accuracy at a minimum scale.

[0108] In some embodiments, the suspension dynamics system construction method further includes establishing a tire model, wherein the tire model adopts a magic formula tire model, and the tire model is used to obtain tire force according to tire positioning parameters.

[0109] In some embodiments, the kinematic pairs in the actual vehicle multi-body system are simplified by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, including:

[0110] The bushing connecting the upper end of the shock absorber to the vehicle frame is simplified to a three-degree-of-freedom ball joint, ignoring the degree of freedom in one direction. A first auxiliary rigid body is introduced, with two mutually perpendicular revolute pairs at both ends, thereby replacing the ball joint with two single-degree-of-freedom revolute pairs.

[0111] There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body is introduced, thereby replacing the cylindrical pair with a rotational pair and a translation pair.

[0112] The suspension dynamics system in any of the above embodiments can be established through the suspension dynamics system construction method of the present application.

[0113] In some embodiments of the present application, a method for dynamic simulation of a suspension system is further provided. The method for dynamic simulation uses the suspension dynamic system described in any of the above embodiments for simulation.

[0114] The present invention establishes a recursive multi-body model based on quasi-speed through quasi-speed characterization and suspension recursive multi-body dynamics model. At the same time, the present invention establishes a data-driven K / C model (i.e., a neural network model). The present invention integrates the trained neural network model with the recursive multi-body model to obtain a hybrid model with high real-time performance and high precision. During the calculation process, the present invention inputs the tire force calculated by the tire model at each step into the neural network model to obtain the change in tire alignment parameters caused by the wheel force in the next step. At the same time, the K / C characteristic influence of the tire alignment parameters is superimposed with the tire alignment parameters caused by the suspension motion constraints to obtain the corrected tire alignment parameters in the current simulation process and then calculate the tire force to obtain a simulation result with higher precision.

[0115] refer to Figure 2In an embodiment of the present application, the initial tire force can be determined based on empirical data, and an initial value of the tire force can be set as input. Through this initial value, the neural network model and the recursive multi-body model are input to obtain tire positioning parameters, and the tire positioning parameters of this time step are used to obtain new tire forces through the MF model. The new tire forces can be directly input into the recursive multi-body model to obtain the posture change of the entire vehicle, that is, to obtain the posture change parameters of the vehicle, including the displacement of the center of mass and the posture angle of the vehicle. In some embodiments of the present application, each time a new tire force (i.e., the tire force of the next time step) can be obtained from the tire positioning parameters of the previous time step through the MF model, it can be used as input to input the neural network model and the recursive multi-body model again to obtain tire positioning parameters, and this cycle is repeated. After the recursive multi-body model calculates the posture change of the entire vehicle, the next cycle begins again until the simulation ends.

[0116] In some embodiments of the present application, t equals t_end as a simulation end condition, where t_end is a preset time threshold.

[0117] It will be easily understood by those skilled in the art that the above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A suspension dynamics system, characterized in that: The suspension dynamics system includes: A recursive multi-body model, which simplifies the kinematic pairs in an actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation; and by shearing the vehicle multi-body system at preset positions, transforms the closed-loop topology of the vehicle multi-body system into an open-loop structure containing multiple branch chains, and achieves equivalence between the open-loop structure and the original closed-loop topology by adding constraint equations; The recursive multi-body model is a multi-body dynamics system described in the form of ordinary differential equations. The multi-body dynamics system described in the form of ordinary differential equations is obtained by the following method: Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates; The dependent hinge coordinate z d With independent hinge coordinate z i It indicates that the dynamic system is described by the body quasi-velocity and independent hinge coordinates; The multi-body dynamics system described in the form of differential algebraic equations is transformed into a multi-body dynamics system described in the form of ordinary differential equations; The recursive multi-body model is used to obtain the K characteristic parameters of the suspension according to the tire force; The suspension dynamics system further includes: a neural network model, the neural network model being used to obtain C characteristic parameters of the suspension based on tire forces; The recursive multi-body model is also used to obtain the vehicle's posture change parameters based on the K characteristic parameters and the C characteristic parameters.

2. The suspension dynamics system according to claim 1, characterized in that: The suspension dynamics system further includes a tire model, wherein the tire model is used to obtain tire force according to tire alignment parameters, wherein the tire alignment parameters are obtained by summing the K characteristic parameter and the C characteristic parameter.

3. The suspension dynamics system according to claim 1, characterized in that: The recursive multi-body model simplifies the kinematic pairs in the actual vehicle multi-body system by introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, including: The bushing connecting the upper end of the shock absorber to the vehicle frame is simplified to a three-degree-of-freedom ball joint, ignoring the degree of freedom in one direction. A first auxiliary rigid body is introduced, with two mutually perpendicular revolute pairs at both ends, thereby replacing the ball joint with two single-degree-of-freedom revolute pairs. There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body is introduced, thereby replacing the cylindrical pair with a rotational pair and a translation pair.

4. The suspension dynamics system according to claim 1, characterized in that: The neural network model includes a fully connected neural network or a recurrent neural network; For the fully connected neural network, the input is a column vector consisting of the longitudinal force, lateral force, and self-aligning moment of the tire at the current moment and several previous moments, and the output is the tire positioning parameter under the coupled working condition; For the recurrent neural network, the input is a column vector consisting of the longitudinal force, lateral force, and self-aligning torque of the tire at the current moment, and the output is the tire positioning parameters under the coupled working condition.

5. The suspension dynamics system according to claim 4, characterized in that: The suspension dynamics system adopts the Adams-Bashforth-Moulton algorithm as a benchmark algorithm to improve the real-time performance of simulation calculations.

6. A method for constructing a suspension dynamics system, characterized in that: The suspension dynamics system construction method includes: establishing a recursive multi-body model and a neural network model; Building a recursive multi-body model involves: By introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, the kinematic pairs in the actual vehicle multi-body system are simplified; and by shearing the preset positions of the vehicle multi-body system, the closed-loop topology structure of the vehicle multi-body system is transformed into an open-loop structure containing multiple branch chains, and the equivalence of the open-loop structure to the original closed-loop topology structure is achieved by adding constraint equations; Converting the dynamic system described entirely by quasi-velocity into a dynamic system described jointly by the vehicle body quasi-velocity and hinge coordinates; wherein the hinge coordinates are divided into independent hinge coordinates and dependent hinge coordinates; The dependent hinge coordinate z d With independent hinge coordinate z i It indicates that the dynamic system is described by the body quasi-velocity and independent hinge coordinates; Transform the multi-body dynamics system described in the form of differential algebraic equations into a multi-body dynamics system described in the form of ordinary differential equations; Wherein, the neural network model includes a fully connected neural network or a recurrent neural network.

7. The method for constructing a suspension dynamics system according to claim 6, wherein: Establishing the fully connected neural network includes: performing hysteresis removal processing on data containing hysteresis curves, using the hysteresis-removed data for training, adjusting network parameters to prevent overfitting, and selecting the network structure and network parameters with the best fitting effect; Establishing the recurrent neural network includes: retaining curve hysteresis information in experimental or simulated data, optimizing the network structure and scale, and obtaining a preset simulation accuracy at a minimum scale.

8. The method for constructing a suspension dynamics system according to claim 7, wherein: The suspension dynamics system construction method further includes establishing a tire model, wherein the tire model adopts a magic formula tire model, and the tire model is used to obtain tire force according to tire positioning parameters.

9. The method for constructing a suspension dynamics system according to claim 8, wherein: By introducing auxiliary rigid bodies and corresponding kinematic pairs for equivalent transformation, the kinematic pairs in the actual vehicle multi-body system are simplified, including: The bushing connecting the upper end of the shock absorber to the vehicle frame is simplified to a three-degree-of-freedom ball joint, ignoring the degree of freedom in one direction. A first auxiliary rigid body is introduced, with two mutually perpendicular revolute pairs at both ends, thereby replacing the ball joint with two single-degree-of-freedom revolute pairs. There is a cylindrical pair between the sleeve at the upper end of the shock absorber and the steering knuckle. The cylindrical pair is decomposed into one rotational degree of freedom and one translational degree of freedom, and a second auxiliary rigid body is introduced, thereby replacing the cylindrical pair with a rotational pair and a translation pair.

10. A method for dynamic simulation of a suspension system, characterized in that: The dynamic simulation method uses the suspension dynamics system described in any one of claims 1 to 5 for simulation.

Images