Semi-active suspension vibration control method and system based on preview fusion and game decision

Abstract

The invention provides a semi-active suspension vibration control method and system based on preview fusion and game decision, and the method comprises the following steps: obtaining pavement preview information through a forward sensor, carrying out the signal processing, and generating a future pavement preview sequence in real time; establishing a two-degree-of-freedom 1 / 4 whole vehicle vertical dynamical model; under a model prediction control framework, participants of a non-cooperative Nash game model are defined, a corresponding cost function is designed, a damping coefficient is discretized by adopting mixed integer programming, Nash equilibrium is solved by adopting a branch and bound method, and an optimal damping coefficient is obtained; mapping the optimal damping coefficient into optimal damping force meeting constraint conditions; hysteresis force compensation is conducted through a hysteresis model, and target control current is obtained through damping force obtained after hysteresis compensation according to the mapping relation between the damping force of the magnetorheological damper and the control current; and outputting the target damping force according to the instruction of the target control current. According to the invention, autonomous balance adjustment of multiple control targets is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent control of automobile suspensions, and particularly relates to a semi-active suspension vibration control method and system based on pre-look fusion and game decision. BACKGROUND

[0002] During vehicle driving, uneven road excitation is the main factor causing suspension vibration and affecting ride comfort. Traditional passive suspensions cannot adapt to complex and changeable road conditions due to fixed damping parameters. Existing semi-active suspension control methods usually need to manually set control weights of different targets in order to simultaneously meet multiple target requirements such as ride comfort and handling stability. This method has large differences in control effect under different working conditions and lacks self-adaptive ability.

[0003] Although the pre-look control technology can obtain the front road information through sensors such as cameras or laser radars, in the existing method, the information is only used as feedforward compensation, and cannot be deeply fused with the decision-making process of suspension vibration control, resulting in a lag in the damping force response. Under random road excitation, the control effect is still different from that of an ideal active suspension. SUMMARY

[0004] In view of the deficiencies in the prior art, the present application provides a semi-active suspension vibration control method and system based on pre-look fusion and game decision, which introduces the non-cooperative game theory into the pre-look control of the semi-active suspension, regards the ride comfort, handling stability and energy economy of the vehicle as control targets of different participants that can be independently decided, and through sensor pre-look, the road disturbance is included in the game decision in advance, and through Nash equilibrium dynamic optimization, the self-balanced adjustment of multiple control targets is realized, thereby providing a new method for semi-active suspension control of the vehicle and having a wide application prospect.

[0005] The present application achieves the above technical objectives through the following technical means.

[0006] A semi-active suspension vibration control method based on pre-look fusion and game decision, comprising the following steps:

[0007] Obtaining road pre-look information through a forward sensor and performing signal processing to generate a future road pre-look sequence in real time;

[0008] establishing a two-degree-of-freedom 1 / 4 vehicle vertical dynamics model including a sprung mass m s , a non-sprung mass m u , a suspension spring stiffness k s , a tire stiffness k t , and an adjustable damping shock absorber;

[0009] In the model predictive control framework, participants of a non-cooperative Nash game model are defined for the ride comfort target, the handling stability target and the energy economy target, and corresponding cost functions are designed, the damping coefficient is discretized by using mixed integer programming, the Nash equilibrium is solved by using branch and bound method, and the optimal damping coefficient is obtained; the optimal damping coefficient is mapped to the optimal damping force meeting the constraint condition;

[0010] The Bouc-Wen hysteresis model is used for hysteresis force compensation, and the compensated damping force F d,cmd is obtained d,cmd According to the mapping relationship between the damping force of the magneto-rheological damper and the control current, the target control current can be obtained;

[0011] The magneto-rheological damper outputs the target damping force according to the instruction of the target control current, and is used for vertical vibration control of the suspension.

[0012] Further, the future road surface preview sequence is generated in real time, and the specific steps are as follows:

[0013] Road surface information is obtained by fusing a forward binocular camera and a laser radar, the camera is used as a master clock, and time and space synchronous collection of sensor data is realized based on a PTP protocol;

[0014] Through coordinate transformation, all point cloud data is converted from a laser radar coordinate system to a vehicle coordinate system, a random sample consensus algorithm (RANSAC) is used for ground segmentation, and the laser radar point cloud data is subjected to grid processing, random obstacle detection is performed in combination with a YOLOv4-tiny algorithm, and a road surface space elevation z r (x) is generated, wherein x is a longitudinal distance relative to a current position of the vehicle;

[0015] The vehicle speed v(t) and the acceleration a(t) of the automobile collected by the inertial measurement unit are used to obtain the predicted driving speed v(t+iΔt) based on a first-order motion model, and the vehicle longitudinal driving distance s(t) is calculated, wherein i is an accumulated integral index for speed prediction, and Δt is a control period;

[0016] The pitch angle of the automobile during acceleration is corrected, and the pitch angle compensation model is:

[0017]

[0018] wherein L is the distance from the sensor center to the front axle of the automobile, θ pitch is the pitch angle of the automobile, and z r,corrected (x) is the corrected space elevation;

[0019] According to the obtained vehicle longitudinal driving distance s(t) and the predicted driving speed v(t+iΔt), the corrected space elevation zr,corrected (x) converting into the road preview sequence z corresponding to the future time window pre (t+kΔt);

[0020] compensating for the total delay of the road preview sequence z pre (t+kΔt);

[0021] further, the predicted driving speed ;

[0022] the road preview sequence z corresponding to the future time window pre (t+kΔt) is specifically:

[0023] ,

[0024] wherein k is the time step index of the preview sequence, and Δt is the control period;

[0025] the total delay compensation of the road preview sequence z pre (t+kΔt) is specifically as follows:

[0026] ,

[0027] ,

[0028] wherein τ total is the total delay, τ sensor is the sensor acquisition delay, τ process is the data processing delay, and τ act is the actuator response delay.

[0029] Further, the dynamic equation of the two-degree-of-freedom 1 / 4 vehicle vertical dynamics model is:

[0030] ,

[0031] ,

[0032] wherein c s is the equivalent passive damping coefficient of the shock absorber; z s is the vertical displacement of the sprung mass, z u is the vertical displacement of the unsprung mass; z r is the road roughness; F d is the damping force generated by the semi-active suspension magneto-rheological shock absorber; is the vertical velocity of the vehicle body; is the vertical acceleration of the vehicle body; is the vertical velocity of the unsprung mass; is the vertical acceleration of the unsprung mass; k s is the tire stiffness;

[0033] The damping force generated by the semi-active suspension magnetorheological damper in the dynamic equation is expressed as:

[0034] ,

[0035] Where F0 is the maximum saturation damping force of the magnetorheological fluid, in N; β1 is the velocity coefficient, in s / m; β2 is the current coefficient, in 1 / A; b1 is the linear damping coefficient, in Ns / m; and I is the control current, in A.

[0036] The tire stiffness in the dynamic equation can be expressed as:

[0037] ,

[0038] Where, k t0 For tire static stiffness, ε is the hysteresis intensity coefficient, and ε is the velocity influence coefficient, in units of s / m.

[0039] Furthermore, the participants in the non-cooperative Nash game model include:

[0040] The first participant, to minimize the vertical acceleration of the vehicle body. The objective; the second participant, to minimize tire dynamic load. The objective is to minimize the change in damping force; a third participant. For the goal;

[0041] The cost functions of the three participants, denoted as J1, J2, and J3, are as follows:

[0042] ,

[0043] ,

[0044] ;

[0045] In the formula, q a q is the weighting coefficient corresponding to the vertical acceleration of the vehicle body; t r is the weighting coefficient corresponding to the tire dynamic load; Δ The weighting coefficients corresponding to the changes in damping force; F is the vertical acceleration of the vehicle body. dyn For tire dynamic load; Δc cmd This represents the change in damping force.

[0046] Based on road surface preview sequence As a known disturbance, the state equations of the model predictive control framework (MPC) are as follows:

[0047]

[0048] in, Let k be the state vector at time k. The vertical velocity of the vehicle body. The vertical velocity of the unsprung mass; c is the state vector at time k+1; cmd,k Let be the damping coefficient at time k; A, B, and D are discrete state matrices, derived from the established dynamic equations, specifically as follows:

[0049] , , .

[0050] Furthermore, the damping coefficient is discretized using mixed integer programming, and the Nash equilibrium is solved using the branch and bound method to obtain the optimal damping coefficient, specifically:

[0051] The damping coefficient is discretized into finite levels using mixed integer programming. The damping coefficient value of each level is dynamically distributed according to the nonlinear characteristics of the shock absorber, satisfying the following relationship:

[0052]

[0053] Among them, c j c is the damping coefficient value for the j-th gear; min The minimum permissible damping coefficient; c max The maximum allowable damping coefficient; m is the total number of damping stops after discretization of the damping coefficient; It is a non-linear distribution index;

[0054] Given the participant set {participant1, participant2, participant3} and the finite gear strategy set {c1, c2, ..., c...} m Given the cost function set {J1, J2, J3}, the Nash equilibrium point of the non-cooperative game is solved iteratively using the branch and bound method to obtain the optimal damping coefficient, which is then mapped to the optimal damping force. During the solution process, the following constraints must be satisfied: Damping coefficient control c min ≤ c ≤ c max In the formula c min For the minimum damping coefficient, c max This is the maximum damping coefficient.

[0055] Furthermore, the Bouc-Wen hysteresis model was used for hysteresis force compensation to obtain the compensated damping force F. d,cmd The details are as follows:

[0056] The Bouc-Wen model is used to calculate the hysteresis force F d,hyst , and the formula is ,

[0057] Wherein, alpha is the hysteresis stiffness coefficient, beta is the hysteresis damping coefficient, The hysteresis displacement internal variable, The suspension relative velocity;

[0058] Hysteresis displacement internal variable And the suspension relative velocity Respectively:

[0059] ,

[0060] ,

[0061] Wherein, The rate of change of the hysteresis displacement internal variable Eta is the asymmetric attenuation coefficient, delta is the symmetric attenuation coefficient, n is the nonlinear order, and A' is the linear gain coefficient;

[0062] The damping force F after hysteresis compensation d,cmd Is: .

[0063] A semi-active suspension vibration control system based on preview fusion and game decision, comprising a processor and a memory, the memory stores computer readable instructions, when the computer readable instructions are executed by the processor, the steps of the semi-active suspension vibration control method based on preview fusion and game decision are run.

[0064] The beneficial effects of the present application are:

[0065] The semi-active suspension vibration control method based on preview fusion and game decision, by the sensor previewing the front road surface information, introduces the non-cooperative game theory into the semi-active suspension vibration control, improves the automobile ride comfort, and takes into account the automobile steering stability and energy economy, realizes the self-balancing of multiple control targets, and provides a reference for the intelligent control of automobile suspension. BRIEF DESCRIPTION OF DRAWINGS

[0066] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings needed to be used in the embodiment or prior art description will be briefly introduced as follows, and the drawings in the following description are some embodiments of the present application. For those skilled in the art, without creative labor, other drawings can also be obtained according to these drawings.

[0067] Figure 1Flow chart of the semi-active suspension vibration control method based on pre- vision fusion and game decision described in the present application.

[0068] Figure 2 Flow chart of the implementation of data acquisition and pre- vision sequence generation.

[0069] Figure 3 Automobile two-degree-of-freedom vertical dynamics model.

[0070] Figure 4 Time-domain comparison curve of body vertical acceleration obtained by the proposed semi-active suspension vibration control method and the traditional passive suspension control method in the embodiment.

[0071] Figure 5 Time-domain comparison curve of tire dynamic load obtained by the proposed semi-active suspension vibration control method and the traditional passive suspension control method in the embodiment.

[0072] Figure 6 Comparison result of body vertical acceleration RMS value obtained by the proposed semi-active suspension vibration control method and the traditional passive suspension control method in the embodiment.

[0073] Figure 7 Comparison result of tire dynamic load RMS value obtained by the proposed semi-active suspension vibration control method and the traditional passive suspension control method in the embodiment. DETAILED DESCRIPTION

[0074] The embodiments of the present application are described in detail below with reference to the accompanying drawings, wherein the same or similar notations represent the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to explain the present application, and cannot be understood as limiting the present application.

[0075] In the description of the present application, it should be understood that the terms "center", "longitudinal", "transverse", "length", "width", "thickness", "upper", "lower", "axial", "radial", "vertical", "horizontal", "inner", "outer", etc. indicate the orientation or positional relationship based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the devices or elements referred to must have a particular orientation, be constructed and operated in a particular orientation, and therefore cannot be understood as limiting the present application. In addition, the terms "first", "second" are only for the purpose of description, and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined with "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "multiple" is two or more, unless otherwise specifically limited.

[0076] In the present application, unless otherwise explicitly specified and limited, the terms "mounting", "connecting", "connecting", "fixing" and the like should be understood broadly, for example, it can be fixedly connected, or it can be detachably connected, or integrally connected; it can be mechanically connected, or it can be electrically connected; it can be directly connected, or it can be indirectly connected through an intermediate medium, or it can be the communication inside two elements. For those skilled in the art, the specific meaning of the above terms in the present application can be understood according to the specific circumstances.

[0077] As shown in Figure 1 The semi-active suspension vibration control method based on pre-foresight fusion and game decision of the present application comprises the following steps:

[0078] S01, road pre-foresight information acquisition and processing, Figure 2 In order to realize the implementation process of road information collection and output of pre-foresight time sequence, specifically:

[0079] The road information is obtained by installing a forward binocular camera and a laser radar fusion system on the automobile. The binocular camera is used to capture the road texture features, and the laser radar is used to obtain the three-dimensional Euclidean distance from the emission point to the specific point of the front road. At the same time, the vehicle speed v(t) is obtained through the inertial measurement unit IMU (Inertial Measurement Unit), and the relative displacement and attitude change of the vehicle are calculated to obtain the longitudinal driving distance s(t) of the vehicle on the running track. Taking the camera as the master clock, the time and space synchronous collection of sensor data is realized based on PTP (Precision Time Protocol), that is, the microsecond-level synchronization of camera data, laser radar data and IMU data is realized based on PTP.

[0080] Through coordinate transformation, all point cloud data is converted from the laser radar coordinate system to the vehicle coordinate system (the coordinate origin is located at the vehicle mass center position), the random sample consensus algorithm (RANSAC) is used for ground segmentation, and the laser radar point cloud data is grid processed, combined with the YOLOv4-tiny method to detect random obstacles such as road pits or speed bumps, and inject height mutation values at the corresponding point cloud data positions to generate road space height z r (x) that meets the target resolution, wherein x is the longitudinal distance relative to the current position of the vehicle;

[0081] According to the vehicle speed v(t) and acceleration a(t) collected by the IMU, the predicted driving speed v(t+iΔt) is obtained based on a first-order motion model, and the calculation model is: ; and the vehicle longitudinal driving distance s(t) is calculated;

[0082] Where i is the cumulative integral index of vehicle speed prediction, and Δt is the control period.

[0083] The pitch angle of the vehicle during acceleration is corrected, and the pitch angle compensation model is:

[0084]

[0085] Where L is the distance from the sensor center to the front axle of the vehicle, θ pitch is the pitch angle of the vehicle; z r,corrected (x) is the corrected spatial elevation.

[0086] According to the obtained vehicle longitudinal driving distance s(t) and the predicted driving speed v(t+iΔt), the corrected spatial elevation z r,corrected (x) is converted into the road preview sequence z pre (t+kΔt) corresponding to the future time window, and the conversion formula is as follows:

[0087] ,

[0088] Where k is the time step index of the preview sequence, and Δt is the control period.

[0089] The total delay compensation of the road preview sequence z pre (t+kΔt) is performed, and the modified conversion formula is as follows:

[0090]

[0091]

[0092] Where τ total is the total delay, τ sensor is the sensor acquisition delay, τ process is the data processing delay, and τ act is the actuator response delay.

[0093] S02, a simplified two-degree-of-freedom vehicle vertical dynamics model is established.

[0094] As shown in Figure 3 , a two-degree-of-freedom 1 / 4 vehicle vertical dynamics model including sprung mass (i.e. vehicle body) m s , unsprung mass m u , suspension spring stiffness k s , tire stiffness k t , and adjustable damping shock absorber is established, and its dynamics equation is:

[0095]

[0096]

[0097] c s is the equivalent passive damping coefficient of the shock absorber; z s is the vertical displacement of the vehicle body, z u is the vertical displacement of the unsprung mass; z r is the road roughness; F d is the damping force generated by the semi-active suspension magneto-rheological shock absorber; is the vertical velocity of the vehicle body; is the vertical acceleration of the vehicle body; is the vertical velocity of the unsprung mass; is the vertical acceleration of the unsprung mass; k s is the tire stiffness;

[0098] The damping force generated by the semi-active suspension magneto-rheological shock absorber in the above formula can be expressed as:

[0099]

[0100] where F0 is the maximum saturated damping force of the magneto-rheological fluid, with units of N; β1 is the velocity term coefficient, with units of s / m; β2 is the current term coefficient, with units of 1 / A; b1 is the linear damping coefficient, with units of Ns / m; I is the control current, with units of A. The formula for the damping force generated by the semi-active suspension magneto-rheological shock absorber can be understood as the mapping relationship between the magneto-rheological shock absorber damping force and the control current, i.e. .

[0101] The tire stiffness in the above formula can be expressed as:

[0102]

[0103] where k t0 is the tire static stiffness, is the hysteresis effect strength coefficient, and ε is the velocity influence coefficient, with units of s / m.

[0104] S03, non-cooperative Nash game modeling and solving, as follows:

[0105] Under the model predictive control (MPC) framework, a non-cooperative Nash game model containing three participants is defined, denoted as the first participant, the second participant, and the third participant, where:

[0106] The first participant aims to minimize the vertical acceleration of the vehicle body as the target to reduce the vibration of the vehicle body and meet the ride comfort requirements;

[0107] The second participant aims to minimize the tire dynamic load as the target to ensure the wheel grounding and meet the handling stability requirements;

[0108] Third participant, to minimize the damping force variation The target is to reduce the severe hydraulic impact of the shock absorber caused by current mutation, prolong the service life of the suspension, reduce the energy consumption of the shock absorber, and avoid inducing secondary vibration.

[0109] For different control targets, cost functions are designed for the three participants respectively, denoted as J1, J2 and J3, which are as follows:

[0110] ,

[0111] ,

[0112] ;

[0113] In the formula, q a is the weight coefficient corresponding to the body vertical acceleration; q t is the weight coefficient corresponding to the tire dynamic load; r Δ is the weight coefficient corresponding to the damping force variation; is the body vertical acceleration; F dyn is the tire dynamic load; Δc cmd is the damping force variation, and the damping coefficient corresponding to each preview time step is between the minimum damping coefficient and the maximum damping coefficient allowed by the magneto-rheological shock absorber.

[0114] Based on the road preview sequence As a known disturbance, the state equation of MPC is designed:

[0115] ,

[0116] Among them, is the state vector at time k, is the body vertical velocity, is the non-sprung mass vertical velocity; is the state vector at time k+1; c cmd,k is the damping coefficient at time k; A, B and D are discrete state matrices, which are derived from the established dynamics equation, and are as follows:

[0117] , , .

[0118] The damping coefficient is discretized into a finite gear position by using mixed integer programming, and the damping coefficient value of each gear position is dynamically distributed according to the nonlinear characteristics of the shock absorber, which satisfies the following relationship:

[0119]

[0120] Among them, cj is the damping coefficient value of the jth gear; c min is the minimum damping coefficient allowed; c max is the maximum damping coefficient allowed; m is the total number of gears after the damping coefficient is discretized, and its value is not less than 8; is a nonlinear distribution index, and its value is between 1.5 and 2.0.

[0121] According to the participant set {participant 1, participant 2, participant 3}, the limited gear strategy set {c1, c2, …, c m}, and the cost function set {J1, J2, J3}, the Nash equilibrium point of the non-cooperative game is iteratively solved by using the branch and bound method, the optimal damping coefficient is obtained, and it is mapped to the optimal damping force , and the following constraint conditions are controlled during the solving process: , the damping coefficient control c min ≤ c ≤ c max ;

[0122] S04, determine the hysteresis compensation and the target control current;

[0123] Considering that the magneto-rheological damper has strong hysteresis characteristics in the dynamic damping force response, the Bouc-Wen hysteresis model is used to calculate the hysteresis force F d,hyst , and the specific formula is as follows:

[0124]

[0125] wherein, α is the hysteresis stiffness coefficient, β is the hysteresis damping coefficient, is the hysteresis displacement internal variable, is the suspension relative speed.

[0126] The hysteresis displacement internal variable and the suspension relative speed are respectively:

[0127]

[0128]

[0129] wherein, is the change rate of the hysteresis displacement internal variable , η is the asymmetric attenuation coefficient, δ is the symmetric attenuation coefficient, n is the nonlinear order, and A’ is the linear gain coefficient.

[0130] The damping force F d,cmd after hysteresis compensation is:

[0131] Damping force F after hysteresis compensation d,cmd Based on the mapping relationship between the damping force of the magnetorheological damper and the control current, the target control current can be obtained.

[0132] S05, Suspension vibration control activated;

[0133] The magnetorheological damper outputs a hysteresis-compensated damping force F based on the received target control current command. d,cmd (i.e., target damping force), to realize the output of target damping force and control of vehicle vertical vibration.

[0134] Example:

[0135] To illustrate the advantages of the control method proposed in this invention in improving vehicle ride comfort and enhancing handling stability, a double lane change test was conducted according to the ISO 3888-2 standard. This test fully considers the vehicle pitch and roll caused by the shift of the vehicle's center of gravity.

[0136] The vehicle speed was controlled at 80 km / h, and the changes in vertical dynamic load during two rapid turns were analyzed. Furthermore, to simulate this effect in the vehicle's vertical dynamics model, the lateral acceleration information of the double lane change scenario was converted into an equivalent vertical excitation. In this embodiment, the double lane change process is approximated as two consecutive "half-sine wave" shaped road surfaces. Specifically: from t=2.0 to 3.0 s, there is an upward half-sine wave bulge, simulating the first turn; from t=4.5 to 5.5 s, there is a downward half-sine wave depression, simulating the second reverse turn. The amplitude of the road surface input is ±5 cm, and the frequency is 1 Hz.

[0137] and Figure 3 The corresponding parameters for the two-degree-of-freedom vertical dynamics model of the car are: m s =320kg, m u =40kg, k s =40000 N / m, k t =200000N / m, c s =750 N·s / m.

[0138] According to the embodiments described above, the semi-active suspension control method based on preview fusion and game theory decision-making proposed in this invention can quickly identify road surface information ahead and generate high-precision elevation information through a camera and lidar fusion system. Combined with the current vehicle speed and position, and a spatiotemporal data conversion model, a road surface preview sequence for future time windows can be generated.

[0139] The time-domain comparison curves of vehicle vertical acceleration and tire dynamic load obtained by using traditional passive suspension and the semi-active suspension control method proposed in this invention are as follows: Figure 4and Figure 5 The body vertical acceleration and tire dynamic load RMS values corresponding to the two suspension control methods are extracted, and the results are shown in Figures 6 and 7, respectively. Figure 6 and Figure 7 It can be seen that the semi-active suspension control method based on preview fusion and game decision proposed in the application can effectively reduce the body vertical acceleration and tire dynamic load, improve the ride comfort of the automobile, and take into account the handling stability of the automobile. In terms of energy consumption economy, the method proposed in the application also achieves better energy consumption effect.

[0140] The semi-active suspension vibration control method based on preview fusion and game decision proposed in the application obtains Nash equilibrium based on the acquisition of road preview information and the establishment of a non-cooperative game model, improves the ride comfort of the automobile, takes into account the handling stability and energy consumption economy of the automobile, realizes dynamic coordination of multiple control objectives, and provides a reference for intelligent suspension control of the automobile.

[0141] The semi-active suspension vibration control system based on preview fusion and game decision proposed in the application comprises a processor and a memory, the memory stores computer readable instructions, and when the computer readable instructions are executed by the processor, the steps in the semi-active suspension vibration control method based on preview fusion and game decision are run.

[0142] It should be understood that although the present specification is described in terms of various embodiments, not every embodiment contains only one independent technical solution, and the description manner of the specification is only for the sake of clarity, and those skilled in the art should consider the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other implementation manners that those skilled in the art can understand.

[0143] The series of detailed descriptions listed above are only specific descriptions of feasible embodiments of the application, and are not used to limit the protection scope of the application, and equivalent embodiments or changes made without departing from the spirit of the application should be included in the protection scope of the application.

Claims

1. A semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making, characterized in that, Includes the following steps: By acquiring road surface preview information through forward sensors and performing signal processing, a future road surface preview sequence can be generated in real time. Establish a system including the sprung mass m s Unsprung mass m u Suspension spring stiffness k s Tire stiffness k t A two-degree-of-freedom, 1 / 4-degree-of-freedom vertical dynamics model of the whole vehicle with adjustable damping shock absorbers; Under the model predictive control framework, participants in a non-cooperative Nash game model are defined for ride comfort, handling stability, and energy economy objectives, and corresponding cost functions are designed. The damping coefficient is discretized using mixed integer programming, and the Nash equilibrium is solved using the branch and bound method to obtain the optimal damping coefficient. The optimal damping coefficient is then mapped to the optimal damping force that satisfies the constraints. The Bouc-Wen hysteresis model was used for hysteresis force compensation to obtain the compensated damping force F. d,cmd Damping force F after hysteresis compensation d,cmd The target control current is obtained based on the mapping relationship between the damping force of the magnetorheological damper and the control current; The magnetorheological damper outputs a target damping force according to the command of the target control current, which is used for vertical vibration control of the suspension.

2. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 1, characterized in that, The specific steps for generating future road surface preview sequences in real time are as follows: Road surface information is obtained by fusing forward-facing binocular cameras and lidar, and the camera is used as the master clock to achieve spatiotemporal synchronous acquisition of sensor data based on the PTP protocol. Through coordinate transformation, all point cloud data is converted from the LiDAR coordinate system to the vehicle coordinate system. Random Sample Consensus Algorithm (RANSAC) is used for ground segmentation, and the LiDAR point cloud data is simultaneously meshed. The YOLOv4-tiny algorithm is then used for random obstacle detection to generate the road surface spatial elevation z. r (x), where x is the longitudinal distance relative to the vehicle's current position; The vehicle speed v(t) and acceleration a(t) are collected by the inertial measurement unit. Based on the first-order motion model, the predicted driving speed v(t+iΔt) is obtained, and the longitudinal driving distance s(t) of the vehicle is calculated. Where i is the cumulative integral index when predicting the vehicle speed, and Δt is the control period. To correct the pitch angle during vehicle acceleration, the pitch angle compensation model is as follows: ， Where L is the distance from the center of the sensor to the front axle of the car, and θ pitch The pitch angle of the car; z r,corrected (x) represents the corrected spatial elevation; Based on the obtained longitudinal travel distance s(t) and the predicted travel speed v(t+iΔt), the corrected spatial elevation z is... r,corrected (x) is converted into the road surface preview sequence z corresponding to the future time window. pre (t+kΔt); For the road surface preview sequence z pre (t+kΔt) is used to compensate for the total delay.

3. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 2, characterized in that, Predicted driving speed ; The road surface prediction sequence z corresponding to the future time window pre (t+kΔt) specifically refers to: ， In the formula: k is the time step index of the pre-aiming sequence, and Δt is the control period; Road surface preview sequence z pre The total delay compensation of (t+kΔt) is as follows: ， ， Where, τ total For the total delay, τ sensor For sensor acquisition delay, τ process For data processing latency, τ act This is due to the actuator response delay.

4. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 1, characterized in that, The dynamic equations of the two-degree-of-freedom 1 / 4 vehicle vertical dynamics model are as follows: ， ， In the formula: c s z is the equivalent passive damping coefficient of the vibration damper. s Z represents the vertical displacement of the spring-loaded mass. u For unsprung mass vertical displacement; z r For road surface unevenness; F d The damping force generated by the magnetorheological damper of the semi-active suspension; The vertical velocity of the vehicle body; The vertical acceleration of the vehicle body; The vertical velocity of the unsprung mass; k is the vertical acceleration of the unsprung mass. s For tire stiffness; The damping force generated by the semi-active suspension magnetorheological damper in the dynamic equation is expressed as: ， Where F0 is the maximum saturation damping force of the magnetorheological fluid, in N; β1 is the velocity coefficient, in s / m; β2 is the current coefficient, in 1 / A; b1 is the linear damping coefficient, in Ns / m; and I is the control current, in A. The tire stiffness in the dynamic equation can be expressed as: ， Where, k t0 For tire static stiffness, ε is the hysteresis intensity coefficient, and ε is the velocity influence coefficient, in units of s / m.

5. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 1, characterized in that, The participants in a non-cooperative Nash game model include: The first participant, to minimize the vertical acceleration of the vehicle body. The objective; the second participant, to minimize tire dynamic load. The objective is to minimize the change in damping force; a third participant. For the goal; The cost functions of the three participants, denoted as J1, J2, and J3, are as follows: ， ， ； In the formula, q a q is the weighting coefficient corresponding to the vertical acceleration of the vehicle body; t r is the weighting coefficient corresponding to the tire dynamic load; Δ The weighting coefficients corresponding to the changes in damping force; F is the vertical acceleration of the vehicle body. dyn For tire dynamic load; Δc cmd This represents the change in damping force. Based on road surface preview sequence As a known disturbance, the state equations of the model predictive control framework (MPC) are as follows: ， in, Let k be the state vector at time k. The vertical velocity of the vehicle body. The vertical velocity of the unsprung mass; c is the state vector at time k+1; cmd,k Let be the damping coefficient at time k; A, B, and D are discrete state matrices, derived from the established dynamic equations, specifically as follows: ， ， 。 6. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 5, characterized in that, The damping coefficient is discretized using mixed integer programming, and the Nash equilibrium is solved using the branch and bound method to obtain the optimal damping coefficient, specifically: The damping coefficient is discretized into finite levels using mixed integer programming. The damping coefficient value of each level is dynamically distributed according to the nonlinear characteristics of the shock absorber, satisfying the following relationship: ， Among them, c j c is the damping coefficient value for the j-th gear; min The minimum permissible damping coefficient; c max The maximum allowable damping coefficient; m is the total number of damping stops after discretization of the damping coefficient; It is a non-linear distribution index; Given the participant set {participant1, participant2, participant3} and the finite gear strategy set {c1, c2, ..., c...} m Given the cost function set {J1, J2, J3}, the Nash equilibrium point of the non-cooperative game is solved iteratively using the branch and bound method to obtain the optimal damping coefficient, which is then mapped to the optimal damping force. During the solution process, the following constraints must be satisfied: Damping coefficient control c min ≤c≤c max In the formula c min For the minimum damping coefficient, c max This is the maximum damping coefficient.

7. The semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making according to claim 1, characterized in that, The Bouc-Wen hysteresis model was used for hysteresis force compensation to obtain the compensated damping force F. d,cmd The details are as follows: The Bouc-Wen model is used to calculate the hysteresis force F. d,hyst The formula is , Where α is the hysteretic stiffness coefficient and β is the hysteretic damping coefficient. For hysteretic displacement internal variables, This refers to the relative speed of the suspension. Hysteresis displacement internal variable Relative speed with suspension They are respectively: ， ， in, Internal variables of hysteretic displacement The rate of change, η is the asymmetric attenuation coefficient, δ is the symmetric attenuation coefficient, n is the nonlinear order, and A' is the linear gain coefficient; Damping force F after hysteresis compensation d,cmd for: .

8. A semi-active suspension vibration control system based on pre-aiming fusion and game theory decision-making, characterized in that, It includes a processor and a memory, the memory storing computer-readable instructions, which, when executed by the processor, perform the steps of the semi-active suspension vibration control method based on pre-aiming fusion and game theory decision-making as described in any one of claims 1-7.

Images