Suspension control method and device based on adhesion feasible region constraint

Abstract

This application discloses a suspension control method and apparatus based on adhesion feasible region constraints, relating to the field of vehicle chassis electronic control technology. The suspension control method based on adhesion feasible region constraints includes: monitoring the vehicle's operating state to obtain the vehicle's working state information; determining the adhesion feasible region constraints based on the working state information and predicted parameter information; constructing a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and adhesion feasible region constraints as the objective function, and determining the front and rear axle actions when the objective function satisfies the constraint conditions; and implementing suspension control based on the front and rear axle actions. This method, by converting adhesion capability into adhesion feasible region constraints and introducing them into model predictive control, can improve the accuracy of suspension control and enhance vehicle driving stability.

Description

Technical Field

[0001] This application relates to the field of vehicle chassis electronic control technology, and in particular to a suspension control method and device based on attachment feasible domain constraints. Background Technology

[0002] Existing suspension model predictive control typically establishes a quarter-vehicle two-degree-of-freedom model, using the vertical displacement of the vehicle body and the vertical displacement of the unsprung mass as state variables. The road surface height sequence measured by the forward-looking sensor is introduced into the prediction equation as a known disturbance. The objective function is a weighted quadratic form of the sprung vertical acceleration, suspension dynamic deflection, and tire dynamic deformation. The maximum actuator force and the maximum suspension compression stroke are used as inequality constraints. The action force sequence in the future prediction time domain is solved through rolling optimization, and only the action force at the current moment is executed.

[0003] In related technologies, the dimensions of the pre-aiming information are incomplete, and the pre-aiming amount is only the geometric elevation of the road. This can easily lead to problems such as tires slipping or locking up, resulting in insufficient vehicle stability. Summary of the Invention

[0004] This application provides a suspension control method and apparatus based on attached feasible domain constraints, which can improve the accuracy of suspension control and enhance vehicle driving stability.

[0005] In a first aspect, embodiments of this application provide a suspension control method based on attachment feasible region constraints, the method comprising: The vehicle's operating status is monitored to obtain the vehicle's working status information, which includes the vehicle's tire normal load, slip ratio, and coefficient of adhesion. Based on the working status information and prediction parameter information, the adhesion feasible domain constraint is determined; the prediction parameter information includes the maximum adhesion utilization rate, the longitudinal force demand sequence in the prediction time domain, and the forward road surface elevation prediction sequence. Under constraints of working power amplitude, working power rate of change, suspension travel, tire contact with the ground, and the adhesion feasible region, a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and working power energy consumption is constructed as the objective function. The working power of the front and rear axles when the objective function satisfies the constraints is then determined. The suspension control of the vehicle is implemented based on the power generated by the front and rear axles.

[0006] In the above embodiments, the method includes: monitoring the vehicle's operating state to obtain the vehicle's working state information; determining the adhesion feasible region constraint based on the working state information and prediction parameter information; constructing a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and adhesion feasible region constraint as an objective function, and determining the front and rear axle actions when the objective function satisfies the constraint conditions; and implementing suspension control on the vehicle based on the front and rear axle actions. This method, by converting adhesion capability into an adhesion feasible region constraint and introducing it into model predictive control, can improve the accuracy of suspension control and enhance vehicle driving stability.

[0007] In one possible implementation, determining the attachment feasible region constraint based on the working state information and the prediction parameter information includes: Based on the normal load, the slip ratio, and the adhesion coefficient, feature extraction is performed in a pre-defined set of reference adhesion characteristic information to obtain multiple adhesion curve features. The multiple attachment curve features are weighted and fused based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty. Based on the conservative adhesion parameters and the prediction parameter information, the minimum normal load requirement information and the normal load prediction information of the vehicle's tires are obtained step by step. The adhesion feasibility region constraint is obtained by comparing the minimum normal load requirement information and the normal load prediction information. The conservative adhesion parameters are obtained by the peak adhesion estimate and the estimation uncertainty.

[0008] The method in this embodiment extracts adhesion curve features from the set of working state information and reference adhesion characteristic information, and then performs weighted fusion of multiple adhesion curve features based on corresponding channel weights to obtain peak adhesion estimates and estimation uncertainties. Conservative adhesion parameters are then obtained from the peak adhesion estimates and estimation uncertainties, and finally, adhesion feasible region constraints are constructed by combining prediction parameter information. This provides a determination mechanism from adhesion features to adhesion feasible region constraints. By performing weighted fusion of multiple adhesion curve features based on corresponding channel weights, the accuracy of suspension control can be improved and the stability of vehicle driving can be enhanced.

[0009] In one possible implementation, feature extraction is performed on a pre-defined set of reference adhesion characteristic information based on the normal load, the slip ratio, and the adhesion coefficient to obtain multiple adhesion curve features, including: Based on the normal load, the slip ratio, and the adhesion coefficient, the pre-set set of adhesion characteristic reference curves is matched with the working conditions and the effective domain is filtered to obtain a subset of adhesion characteristic reference curves that match the current working state, and the effective slip ratio range of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves is determined. Based on the subset of adhesion characteristic reference curves and the effective slip ratio range, the adhesion coefficient of the current working state is compared with the adhesion coefficient of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the same normal load and the same slip ratio to obtain multiple adhesion curve features; the multiple adhesion curve features include at least two of the following: point value features, slip domain cumulative features, and local slope features.

[0010] The method of this embodiment obtains a subset of reference curves and an effective slip ratio interval that match the current working state by performing working condition matching and effective domain filtering on the reference curve set. Then, it extracts at least two types of features from point value features, slip domain cumulative features and local slope features within the subset and effective interval. This provides a specific extraction mechanism for attached curve features, which can effectively weaken the insufficient reliability of a single feature under specific working conditions, improve the accuracy of suspension control, and enhance the stability of vehicle driving.

[0011] In one possible implementation, the working state information further includes longitudinal force; the weighted fusion operation of the multiple adhesion curve features based on corresponding channel weights to obtain the peak adhesion estimate and the estimation uncertainty includes: The point value features, the slip domain cumulative features, and the local slope features are mapped to the same peak attachment space to obtain the point value channel peak attachment candidate value, the cumulative channel peak attachment candidate value, and the slope channel peak attachment candidate value. For each channel, the total variance of the channel is constructed based on the sensitivity Jacobian matrix of the observation vector to the peak attachment candidate value of the channel, the observation noise covariance matrix, the variance of the matching residual between the channel and the attachment characteristic reference curve, and the candidate dispersion; the observation vector is a three-dimensional vector composed of the slip ratio, the longitudinal force, and the normal load. Inverse variance fusion weights are constructed based on the total variance of each channel, and these weights are used as the channel weights. Based on the channel weights, the peak attachment candidate values ​​of each channel are weighted and fused to obtain the peak attachment estimate. The estimation uncertainty is obtained based on the channel weights, the total variance of each channel, and the difference between the peak attachment candidate value and the peak attachment estimate of each channel. The candidate dispersion is determined by the weighted dispersion of the candidate values ​​obtained from each attachment characteristic reference curve within the same channel relative to the channel estimate of that channel.

[0012] The method in this embodiment maps three types of features to the same peak attachment space, constructs the total variance of each channel according to the sensitivity Jacobian matrix, the observation noise covariance matrix, the matching residual variance and the candidate dispersion, and then performs weighted fusion with inverse variance fusion weights to obtain the estimated uncertainty. This provides a multi-channel adaptive fusion and uncertainty propagation mechanism, which can improve the accuracy of suspension control and enhance the stability of vehicle driving.

[0013] In one possible implementation, the point value channel, the cumulative channel, and the slope channel are each equipped with a gating function; the step of constructing inverse variance fusion weights based on the total variance of each channel as the channel weights includes: The inverse variance of each channel is obtained by performing the inverse operation on the total variance of each channel. The channel weights are obtained based on the inverse variance of each channel and the corresponding gating function value.

[0014] The method in this embodiment determines the channel weights by using a gating function and inverse variance, so that invalid channels do not participate in the fusion and reliable channels obtain higher weights. This provides a specific mechanism for determining channel weights, which can improve the accuracy of suspension control and enhance the stability of vehicle driving.

[0015] In one possible implementation, the step of obtaining step-by-step minimum normal load requirement information and tire normal load prediction information of the vehicle based on conservative adhesion parameters and the prediction parameter information, and obtaining adhesion feasibility region constraints by comparing the minimum normal load requirement information and the normal load prediction information, includes: Receive the longitudinal force demand sequence in the predicted time domain given by the longitudinal control system, and allocate the longitudinal force demand sequence to the front axle and rear axle according to the drive arrangement to obtain the longitudinal force demand of the target axle; Based on the conservative adhesion parameters, the maximum adhesion utilization rate, and the target axis longitudinal force requirement, the minimum normal load requirement information for each prediction step is obtained; The normal load prediction information on the same time axis is predicted by the half-vehicle model and the forward road elevation prediction sequence, and the normal load prediction information is compared with the minimum normal load requirement information step by step. When the predicted normal load information is not less than the minimum normal load requirement information, the prediction step is determined to be feasible; otherwise, the prediction step is determined to be infeasible. The feasibility determination of each prediction step constitutes the attached feasible domain constraint.

[0016] The method in this embodiment distributes the longitudinal force demand to the front and rear axles according to the drive arrangement, obtains the minimum normal load demand step by step from the conservative adhesion parameters and the maximum adhesion utilization rate, and compares it step by step with the normal load predicted by the half-vehicle model. This provides a specific construction mechanism for the adhesion feasible domain constraint, which can improve the accuracy of suspension control and improve the stability of vehicle driving.

[0017] In one possible implementation, determining the front and rear axle forces when the objective function satisfies the constraints includes: Based on the Laguerre basis function matrix, a quadratic programming solution is performed on the objective function and constraints to obtain the front and rear axis driving forces when the objective function satisfies the constraints.

[0018] The method in this embodiment, by performing quadratic programming on the objective function and constraints based on the Laguerre basis function matrix, compresses the action sequence in the prediction time domain into a few Laguerre coefficients, providing a specific solution mechanism for model predictive control, which can improve the accuracy of suspension control and enhance the stability of vehicle driving.

[0019] In one possible implementation, the suspension control of the vehicle based on the power applied to the front and rear axles includes: Determine whether the attachment feasible region constraint is feasible; If the result of the judgment is feasible, then the suspension control of the vehicle is implemented according to the power of the front and rear axles; If the result of the judgment is that it is not possible, then the suspension control of the vehicle is implemented according to the target front and rear axles as the driving force; The target's front and rear axles are powered by the following process: Under the constraints of the working force amplitude, the working force change rate, and the suspension travel, the achievable normal load envelope of each axle in the prediction time domain is determined. The maximum allowable longitudinal tire force of each axle is obtained from the achievable normal load envelope, the conservative adhesion parameters, and the maximum adhesion utilization rate. The maximum allowable longitudinal tire force is converted into an upper limit value for the control force parameters, and the upper limit value for the control force parameters is sent to the vehicle's longitudinal control system. By softening the adhesion feasible domain constraint, the front and rear axle working forces are re-determined, and the re-determined front and rear axle working forces are used as the target front and rear axle working forces. The upper limit value for the control force parameters includes an upper limit for the driving torque and an upper limit for the braking force.

[0020] The method of this embodiment solves the envelope under the reachable normal load when the attachment feasible region constraint is not feasible, back-calculates the maximum allowable longitudinal tire force and feeds it back to the longitudinal control system. At the same time, after softening the attachment feasible region constraint, the front and rear axle driving forces are re-determined, providing a longitudinal and vertical cooperative degradation mechanism that can improve the accuracy of suspension control and improve the stability of vehicle driving.

[0021] Secondly, embodiments of this application provide a suspension control device based on attachment feasible region constraints, the device comprising: A status information acquisition unit is used to monitor the operating status of the vehicle and obtain the working status information of the vehicle; the working status information includes the normal load, slip ratio and coefficient of adhesion of the vehicle's tires; The adhesion constraint determination unit is used to determine the adhesion feasible domain constraint based on the working status information and the prediction parameter information; the prediction parameter information includes the maximum adhesion utilization rate, the longitudinal force demand sequence in the prediction time domain, and the forward road surface elevation prediction sequence. The control parameter solving unit is used to construct a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load and working power energy consumption as the objective function under the constraints of working power amplitude, working power change rate, suspension travel, tire contact constraint and the adhesion feasible region constraint, and to determine the front and rear axle working power when the objective function satisfies the constraint conditions. A suspension control actuator is used to implement suspension control on the vehicle based on the power applied to the front and rear axles.

[0022] In one possible implementation, the attachment constraint determination unit is specifically used for: Based on the normal load, the slip ratio, and the adhesion coefficient, feature extraction is performed in a pre-defined set of reference adhesion characteristic information to obtain multiple adhesion curve features. The multiple attachment curve features are weighted and fused based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty. Based on the conservative adhesion parameters and the prediction parameter information, the minimum normal load requirement information and the normal load prediction information of the vehicle's tires are obtained step by step. The adhesion feasibility region constraint is obtained by comparing the minimum normal load requirement information and the normal load prediction information. The conservative adhesion parameters are obtained by the peak adhesion estimate and the estimation uncertainty.

[0023] In one possible implementation, the attachment constraint determination unit is specifically used for: Based on the normal load, the slip ratio, and the adhesion coefficient, the pre-set set of adhesion characteristic reference curves is matched with the working conditions and the effective domain is filtered to obtain a subset of adhesion characteristic reference curves that match the current working state, and the effective slip ratio range of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves is determined. Based on the subset of adhesion characteristic reference curves and the effective slip ratio range, the adhesion coefficient of the current working state is compared with the adhesion coefficient of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the same normal load and the same slip ratio to obtain multiple adhesion curve features; the multiple adhesion curve features include at least two of the following: point value features, slip domain cumulative features, and local slope features.

[0024] In one possible implementation, the working state information further includes longitudinal force; the attachment constraint determination unit is specifically used for: The point value features, the slip domain cumulative features, and the local slope features are mapped to the same peak attachment space to obtain the point value channel peak attachment candidate value, the cumulative channel peak attachment candidate value, and the slope channel peak attachment candidate value. For each channel, the total variance of the channel is constructed based on the sensitivity Jacobian matrix of the observation vector to the peak attachment candidate value of the channel, the observation noise covariance matrix, the variance of the matching residual between the channel and the attachment characteristic reference curve, and the candidate dispersion; the observation vector is a three-dimensional vector composed of the slip ratio, the longitudinal force, and the normal load. Inverse variance fusion weights are constructed based on the total variance of each channel, and these weights are used as the channel weights. Based on the channel weights, the peak attachment candidate values ​​of each channel are weighted and fused to obtain the peak attachment estimate. The estimation uncertainty is obtained based on the channel weights, the total variance of each channel, and the difference between the peak attachment candidate value and the peak attachment estimate of each channel. The candidate dispersion is determined by the weighted dispersion of the candidate values ​​obtained from each attachment characteristic reference curve within the same channel relative to the channel estimate of that channel.

[0025] In one possible implementation, the point value channel, the cumulative channel, and the slope channel are each equipped with a gating function; the attachment constraint determination unit is specifically used for: The inverse variance of each channel is obtained by performing the inverse operation on the total variance of each channel. The channel weights are obtained based on the inverse variance of each channel and the corresponding gating function value.

[0026] In one possible implementation, the attachment constraint determination unit is specifically used for

[0027] Receive the longitudinal force demand sequence in the predicted time domain given by the longitudinal control system, and allocate the longitudinal force demand sequence to the front axle and rear axle according to the drive arrangement to obtain the longitudinal force demand of the target axle; Based on the conservative adhesion parameters, the maximum adhesion utilization rate, and the target axis longitudinal force requirement, the minimum normal load requirement information for each prediction step is obtained; The normal load prediction information on the same time axis is predicted by the half-vehicle model and the forward road elevation prediction sequence, and the normal load prediction information is compared with the minimum normal load requirement information step by step. When the predicted normal load information is not less than the minimum normal load requirement information, the prediction step is determined to be feasible; otherwise, the prediction step is determined to be infeasible. The feasibility determination of each prediction step constitutes the attached feasible domain constraint.

[0028] In one possible implementation, the control parameter solving unit is specifically used for: Based on the Laguerre basis function matrix, a quadratic programming solution is performed on the objective function and constraints to obtain the front and rear axis driving forces when the objective function satisfies the constraints.

[0029] In one possible implementation, the suspension control execution unit is specifically used for: Determine whether the attachment feasible region constraint is feasible; If the result of the judgment is feasible, then the suspension control of the vehicle is implemented according to the power of the front and rear axles; If the result of the judgment is that it is not possible, then the suspension control of the vehicle is implemented according to the target front and rear axles as the driving force; The target's front and rear axles are powered by the following process: Under the constraints of the working force amplitude, the working force change rate, and the suspension travel, the achievable normal load envelope of each axle in the prediction time domain is determined. The maximum allowable longitudinal tire force of each axle is obtained from the achievable normal load envelope, the conservative adhesion parameters, and the maximum adhesion utilization rate. The maximum allowable longitudinal tire force is converted into an upper limit value for the control force parameters, and the upper limit value for the control force parameters is sent to the vehicle's longitudinal control system. By softening the adhesion feasible domain constraint, the front and rear axle working forces are re-determined, and the re-determined front and rear axle working forces are used as the target front and rear axle working forces. The upper limit value for the control force parameters includes an upper limit for the driving torque and an upper limit for the braking force.

[0030] Thirdly, embodiments of this application provide an electronic device including a processor and a memory, wherein the memory stores program code that, when executed by the processor, causes the processor to perform the steps of the suspension control method based on the attached feasible domain constraint described above.

[0031] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the suspension control method based on the attached feasible region constraint as described above.

[0032] Fifthly, embodiments of this application provide a computer program product including computer instructions stored in a computer-readable storage medium; when a processor of a computer device reads the computer instructions from the computer-readable storage medium, the processor executes the computer instructions, causing the computer device to perform the steps of the suspension control method based on the attached feasible domain constraint as described above.

[0033] The technical effects of any of the implementation methods in aspects two through five can be found in the technical effects of the implementation method in aspect one, and will not be repeated here. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 A flowchart illustrating a suspension control method based on attachment feasible region constraints provided in this application embodiment; Figure 2 A schematic diagram of a pitch plane half-vehicle dynamics model for a suspension control method based on attachment feasible domain constraints provided in an embodiment of this application; Figure 3 A schematic diagram illustrating the process of determining the attachment feasible region constraint in a suspension control method based on attachment feasible region constraint provided in an embodiment of this application; Figure 4 A schematic diagram of a reference adhesion characteristic information set for a suspension control method based on adhesion feasible region constraints provided in an embodiment of this application; Figure 5 A schematic diagram illustrating the process of implementing vehicle suspension control based on front and rear axle power, according to an embodiment of this application, for a suspension control method based on attachment feasible domain constraints. Figure 6 A schematic diagram of the structure of a vehicle air conditioning control device provided in an embodiment of this application; Figure 7 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0036] To make the objectives, technical solutions, and advantages of this application clearer, the application will be further described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments in this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0037] The following explanations of some terms used in the embodiments of this application are provided to facilitate understanding by those skilled in the art.

[0038] (1) Peak adhesion coefficient: The peak adhesion coefficient refers to the maximum value of the reference curve of the adhesion characteristics between the tire and the road surface, and is referred to as peak adhesion.

[0039] (2) Utilizing the adhesion coefficient: The adhesion coefficient refers to the ratio of the longitudinal force of the tire to the dynamic normal load of the tire at the current moment, and is abbreviated as the adhesion coefficient.

[0040] (3) Conservative attachment parameter: The conservative attachment parameter refers to the safe value of peak attachment obtained by subtracting a certain multiple of its estimation uncertainty from the peak attachment estimate, which is used by the controller.

[0041] (4) Adhesion utilization rate: Adhesion utilization rate refers to the ratio of longitudinal force demand to the product of conservative adhesion parameters and normal load.

[0042] (5) Attached feasible region: The attached feasible region refers to the set of constraints defined by the minimum normal load requirement for each prediction step.

[0043] (6) Adhesion characteristic reference curve set: The adhesion characteristic reference curve set refers to multiple typical road surface adhesion characteristic reference curves established offline with slip ratio and normal load as independent variables.

[0044] (7) Point value feature: In the embodiments of this application, the point value feature is the ratio of the adhesion coefficient of the current working state to the adhesion coefficient of the j-th reference curve at the current slip ratio and normal load.

[0045] (8) Cumulative characteristics of slip domain: In the embodiments of this application, the cumulative characteristics of slip domain are the ratio of the cumulative area of ​​the measured adhesion characteristics to the cumulative area of ​​the j-th reference curve in the effective slip rate range.

[0046] (9) Local slope feature: In the embodiments of this application, the local slope feature is the ratio of the measured local slope at the current working state to the slope of the j-th reference curve at the current working state.

[0047] (10) Working state: In the embodiments of this application, the working state is the vehicle's operating state corresponding to the normal load of the vehicle's tires, the slip ratio, and the working state information determined by the adhesion coefficient.

[0048] (11) Glide domain: The glide domain refers to the domain of definition with the glide rate as the independent variable, which can be used to distinguish it from the time domain with time as the independent variable.

[0049] (12) Envelope under reachable normal load: Under the condition of only the actuator and suspension travel constraints, predict the maximum value that the minimum normal load can reach in the time domain.

[0050] (13) Incentive retention: Incentive retention refers to the measures taken to actively retain the minimum identifiable incentive when the attachment margin is sufficient, so as to prevent the control action from inhibiting the source of identifiable information.

[0051] (14) Inter-axle delay: Inter-axle delay refers to the time when the rear wheel passes through a certain point on the road surface later than the time when the front wheel passes through the same point. The difference is equal to the wheelbase divided by the vehicle speed.

[0052] This application provides a suspension control method and apparatus based on attachment feasible region constraints. To better understand the technical solution provided in this application, the basic principle of the solution is briefly explained below.

[0053] It should be noted that the terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0054] The technical solutions provided by the embodiments of this application are described below with reference to the accompanying drawings.

[0055] Existing suspension model predictive control typically establishes a quarter-vehicle two-degree-of-freedom model, using the vertical displacement of the vehicle body and the vertical displacement of the unsprung mass as state variables. The road surface height sequence measured by the forward-looking sensor is introduced into the prediction equation as a known disturbance. The objective function is a weighted quadratic form of the sprung vertical acceleration, suspension dynamic deflection, and tire dynamic deformation. The maximum actuator force and the maximum suspension compression stroke are used as inequality constraints. The action force sequence in the future prediction time domain is solved through rolling optimization, and only the action force at the current moment is executed.

[0056] In related technologies, the dimensions of the pre-aiming information are incomplete, and the pre-aiming amount is only the geometric elevation of the road. This can easily lead to problems such as tires slipping or locking up, resulting in insufficient vehicle stability.

[0057] In view of this, embodiments of this application provide a suspension control method and apparatus based on adhesion feasible region constraints. The method includes: monitoring the vehicle's operating state to obtain the vehicle's working state information; determining adhesion feasible region constraints based on the working state information and predicted parameter information; constructing a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and adhesion feasible region constraints as an objective function, and determining the front and rear axle actions when the objective function satisfies the constraints; and implementing suspension control on the vehicle based on the front and rear axle actions. This method, by converting adhesion capability into adhesion feasible region constraints and introducing them into model predictive control, can improve the accuracy of suspension control and enhance vehicle driving stability.

[0058] The preferred embodiments of this application are described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit this application. Furthermore, the embodiments and features described in the embodiments of this application can be combined with each other without conflict.

[0059] The suspension control method based on attachment feasible region constraints provided in the embodiments of this application will be further explained below. A suspension control method based on attachment feasible region constraints, such as... Figure 1 As shown, the method includes the following steps: S101 monitors the vehicle's operating status and obtains the vehicle's working status information, including the vehicle's tire normal load, slip ratio, and coefficient of adhesion.

[0060] In some embodiments of this application, wheel speed, vehicle speed, driving torque, braking torque, inertial measurement unit signals, and suspension displacement signals are collected as basic vehicle information. Then, based on the basic vehicle information, the dynamic normal loads of the front and rear axle tires are estimated online based on the pitch plane half-vehicle dynamics model. Based on the wheel rotation equation, the longitudinal forces of the front and rear axle tires are estimated. Based on the wheel speed and vehicle speed, the slip ratio is determined. The ratio of the longitudinal force to the dynamic normal load is used to obtain the current working state on the adhesion characteristic reference curve using the adhesion coefficient.

[0061] In some embodiments of this application, such as Figure 2 As shown, a four-degree-of-freedom half-vehicle model in the pitch plane is used to determine the dynamic normal loads on the front and rear axle tires. The geometric and inertial parameters of this model are defined as follows: The sprung mass is expressed in kilograms. The moment of inertia of the sprung mass about its center of mass during pitch rotation is expressed in kilograms per square meter. and These are the unsprung masses of the front and rear axles, respectively, in kilograms. and These are the horizontal distances from the vehicle's center of gravity to the front and rear axles, respectively, in meters. This refers to the wheelbase, in meters. The vehicle's center of gravity is the height above the ground, in meters. The total mass of the vehicle is expressed in kilograms; g is the acceleration due to gravity.

[0062] The kinematic variables are defined as follows: The vertical displacement of the center of mass on the spring is expressed in meters, with upward being positive. The vehicle body pitch angle is expressed in radians, with the front of the vehicle tilted upwards as positive. and These are the vertical displacements of the unsprung masses of the front and rear axles, respectively, in meters; and The points represent the road surface elevations at the front and rear wheel contact points, respectively, in meters; one point above the variable represents the first derivative with respect to time, and two points represent the second derivative.

[0063] The component parameters are defined as follows: and These are the front and rear suspension spring stiffnesses, respectively, in Newtons per meter (N / m). and These are the damping coefficients for the front and rear suspensions, respectively, in Newton-seconds per meter. and These are the vertical stiffness of the front and rear tires, respectively, in Newtons per meter (N / m). and These are the vertical damping coefficients for the front and rear tires, respectively, in Newton-seconds per meter.

[0064] The variables of force and torque are defined as follows: and The power is provided by the active suspension of the front and rear axles, respectively, and the unit is Newtons, with the sprung mass and unsprung mass being positive. and These are the longitudinal forces between the front and rear axle tires and the ground, respectively, measured in Newtons, with the direction of vehicle movement as positive. and These are the dynamic normal loads on the front and rear axle tires, respectively, in Newtons. This is the rear axle drive torque, measured in Newton-meters. and These are the angular velocities of the front and rear wheels, respectively, in radians per second. The vehicle speed is expressed in meters per second.

[0065] See Figure 2 This image is a side view of the vehicle. The vehicle body outline is drawn with thin lines, below which is a thick solid line representing the sprung mass rigid beam, with the sprung mass marked on the beam. With pitch moment of inertia The vertical displacement of the center of mass on the spring is marked with a vertical arrow above the beam. The vehicle's pitch angle is marked with an arc between the beam and a horizontal dashed line. The diagram shows three parallel elements connecting the beam and the unsprung mass platform at both the front and back. From left to right, these are symbols for a coil spring, a damper, and a circular actuator, with the suspension spring stiffness indicated. and Suspension damping coefficient and , as a driving force and Below the platform are circular wheels, with their unsprung mass marked. and and vertical displacement of unsprung mass and The wheel's angular velocity is indicated by an arc-shaped arrow inside the wheel. and The driving torque is indicated by an arc arrow on the outer side of the rear wheel. This indicates rear-wheel drive. Springs and dampers are drawn between the wheel and the undulating road surface, and the tire's vertical stiffness is indicated. and Tire vertical damping coefficient and The dynamic normal load is marked with a vertical arrow at the grounding point. and Longitudinal forces are indicated by horizontal arrows. and The center of gravity is marked inside the vehicle body with a circle and a cross symbol, and the vertical arrow below it indicates the weight G. The height of the center of gravity from the ground is marked by a double-sided dimension line. The distance from the center of gravity to the front axle is indicated at the bottom of the diagram by bidirectional dimension lines. Distance from center of gravity to rear axle With wheelbase The speedometer is indicated by a horizontal arrow in the upper left corner. The text in the upper right corner indicates the direction of the rear axle drive and the relationship between the rear wheel road input and the front wheel road input, which is delayed by a wheelbase divided by the vehicle speed.

[0066] In some embodiments of this application, under the assumption of a small pitch angle, the vertical displacement of the upper ends of the front and rear suspensions is: , In the formula, and These represent the vertical displacements of the upper ends of the front and rear suspensions, respectively, in meters.

[0067] The above formula shows that when the front of the car is tilted upward, the upper end of the front suspension moves upward and the upper end of the rear suspension moves downward.

[0068] In some embodiments of this application, the upward vertical forces exerted by the front and rear suspensions on the sprung mass are respectively: , .

[0069] In the formula, and These are the vertical resultant forces acting on the sprung masses by the front and rear suspensions, respectively, in Newtons (N), with upward force being positive. The first item in parentheses is the relative deformation of the suspension, and the second item is the relative deformation rate of the suspension.

[0070] Therefore, we can obtain the system of differential equations for the four degrees of freedom of motion of the half-car: , The first equation is the equation for the vertical translation of the sprung mass; The second equation is the pitch rotation equation of the sprung mass. Its last term is the pitch moment generated by the longitudinal force on the ground around the center of mass. This term enables the model to describe the inter-axle load transfer caused by acceleration squat and braking pitch. Equations three and four are the equations of motion for the front and rear unsprung masses, respectively, and the last two terms are the vertical elastic force and the vertical damping force of the tire, respectively.

[0071] In some embodiments of this application, the generalized displacement vector, control vector, and disturbance vector are defined as follows: , In the formula, It is a four-dimensional generalized displacement vector; It is a two-dimensional control vector; It is a six-dimensional external disturbance vector, whose first four components are the road surface elevation of the front and rear wheels and their rate of change, and the last two components are the longitudinal ground forces of the front and rear axles. The superscript T indicates transpose.

[0072] In some embodiments of this application, substituting the suspension force expression into the system of equations of motion and reducing the terms, we obtain: , In the formula, It is a fourth-order generalized mass matrix. It is a fourth-order generalized damping matrix. It is a fourth-order generalized stiffness matrix. Assign a control matrix to a 4x2 matrix. Assign a matrix to the four rows and six columns of perturbation.

[0073] All elements of each matrix are: , , , , , The above It is a diagonal matrix, and its inverse matrix can be directly obtained from the diagonal elements. and All of them are symmetric matrices, which is a necessary condition for the self-consistency of the model. The first line corresponds to the action of the force on the vertical motion of the spring, the second line corresponds to the action of the force on the pitch motion, and the negative signs in the third and fourth lines indicate that the action of the force on the spring and unsprung mass are equal in magnitude and opposite in direction. The second line indicates that the longitudinal force of the ground enters the system only through the pitch channel, and the third and fourth lines indicate that the road surface elevation and its rate of change enter the unsprung mass only through the tire elasticity and damping.

[0074] In some embodiments of this application, the method includes the construction and discretization process of a state-space model; the construction and discretization process of the state-space model includes: taking an eight-dimensional state vector as a stack of generalized displacements and generalized velocities. , The continuous-time state equations and their coefficient matrices are as follows: , , In the formula, for OK Column zero matrix; It is a fourth-order identity matrix; because It is a diagonal matrix, and its inverse matrix is: , therefore , , Each element can be directly calculated from the component parameters listed in the foregoing embodiments, without any unquantified parameters. (This is for controlling the cycle.) By performing zero-order preserved discretization, we obtain: , , In the formula, For discrete time indices; For matrix exponents; It is the integral variable.

[0075] In some embodiments of this application, the tire normal load output equation is: , , , In the formula, and These are the static axle loads for the front and rear axles, respectively, in Newtons (N).

[0076] Written in matrix form: , , .

[0077] In some embodiments of this application, for rear-wheel drive vehicles, the driving torque is applied only to the rear axle. The longitudinal ground forces on the front and rear axles are estimated by the wheel rotation equation, as follows: , In the formula, the subscript Pick Indicates front axle, take Indicates the rear axle; This is the driving torque of the shaft, measured in Newton-meters, which is always zero for the front axle. This is the braking torque of the shaft, expressed in Newton-meters. This is the moment of inertia of the wheel on this axle, expressed in kilogram square meters. This is the wheel angular acceleration, measured in radians per second squared. The effective rolling radius of the wheel, in meters; is the rolling resistance coefficient, which is dimensionless.

[0078] , In the formula, The signified longitudinal slip ratio is dimensionless, with positive for driving and negative for braking. This is the absolute value of the longitudinal slip ratio; To utilize adhesion, dimensionless; To prevent the lower limit of vehicle speed from being divided by zero, the unit is meters per second; To prevent division by zero, the lower limit of the normal load is given, in Newtons.

[0079] Using real-time estimated dynamic normal loads instead of static axle loads can effectively avoid introducing systematic deviations during active suspension control.

[0080] In some embodiments of this application, the road surface input and the longitudinal force on the ground are set to zero; the dynamic force is taken as a constant value, which is: , Setting all state derivatives to zero, the second-order matrix equation degenerates into: , Will and Substituting the values ​​and solving, the unique solution is: , That is, under constant working force, the sprung mass produces vertical displacement and pitch angle; while the steady-state displacement of the front and rear unsprung masses is always zero; substituting into the output equation, we get that the steady-state increment of the normal load on both axes is zero. Equivalently, the transfer function matrix from the working force to the normal load is a zero matrix at zero frequency. .

[0081] In some embodiments of this application, the four equations of motion are integrated and summed over a time window [0,T] to obtain the impulse constraint of the normal load increment: , In the formula, T is the length of the integration window, in seconds; When the vehicle body and wheels are in the same state at the beginning and end of the window movement, the time integral of the normal load increment is conserved. In the embodiments of this application, the active suspension does not continuously increase the total normal load, and its operation includes: time-dimensional redistribution and inter-axle redistribution.

[0082] In some embodiments of this application, the driving force is decomposed into a buoyancy channel driving force component and a pitch channel driving force component, wherein the buoyancy channel driving force component and the pitch channel driving force component are respectively: , , In the formula, The dynamic component for the floating and sinking channel, measured in Newtons, primarily excites the vertical motion of the spring, at the cost of which is the vertical acceleration of the spring. The pitch channel is represented by a dynamic component, measured in Newton-meters (Nm), which primarily excites the vehicle's pitch motion, at the cost of pitch angular acceleration. These two costs are then weighted and incorporated into the objective function of the suspension control process based on the attachment feasible region constraint.

[0083] In some embodiments of this application, when monitoring the vehicle's operating status and obtaining the vehicle's working status information, it is determined that the vehicle speed estimation is valid and the normal load estimation value is greater than a set lower limit; otherwise, the working status information of the current cycle is marked as invalid.

[0084] S102, determine the adhesion feasible domain constraints based on working status information and prediction parameter information; the prediction parameter information includes the maximum adhesion utilization rate and the longitudinal force demand sequence and the forward road surface elevation prediction sequence in the prediction time domain.

[0085] In some embodiments of this application, the attachment feasible region constraint is determined based on the working state information and prediction parameter information, such as... Figure 3 As shown, this can be achieved through the following steps: S301, based on the normal load, slip ratio and adhesion coefficient, features are extracted from a pre-defined set of adhesion characteristic reference curves to obtain multiple adhesion curve features.

[0086] In some embodiments of this application, the process of extracting features from a pre-defined set of reference adhesion characteristic information based on normal load, slip ratio, and adhesion coefficient to obtain multiple adhesion curve features is achieved through the following steps: Step A01: Based on the normal load, slip ratio, and adhesion coefficient, perform working condition matching and effective domain screening on the pre-set set of adhesion characteristic reference curves to obtain a subset of adhesion characteristic reference curves that match the current working state, and determine the effective slip ratio range of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves.

[0087] Step A02: Based on the subset of adhesion characteristic reference curves and the effective slip ratio range, the adhesion coefficient of the current working state is compared with the adhesion coefficient of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the same normal load and the same slip ratio to obtain multiple adhesion curve features; the multiple adhesion curve features include at least two of the following: point value features, slip domain cumulative features, and local slope features.

[0088] S302 performs a weighted fusion operation on multiple attachment curve features based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty.

[0089] S303: Based on the conservative adhesion parameters and prediction parameters, the minimum normal load requirement information and the normal load prediction information of the vehicle's tires are obtained step by step. The adhesion feasible region constraint is obtained by comparing the minimum normal load requirement information and the normal load prediction information. The conservative adhesion parameters are obtained from the peak adhesion estimate and the estimation uncertainty.

[0090] In some embodiments of this application, the extraction of multiple attachment curve features is referenced. Figure 4 . Figure 4 The reference adhesion characteristic reference curves in the reference adhesion characteristic information set have slip ratio on the x-axis and adhesion coefficient on the y-axis. The figure uses solid lines, long dashed lines, short dashed lines, and dotted lines to draw the adhesion characteristic reference curves for four typical pavements: dry asphalt, wet asphalt, compacted snow, and ice, respectively. The pavement name is labeled at the right end of the curve. Three thin dotted lines are drawn around the dry asphalt curve to represent the same family of pavement curves caused by changes in normal load, labeled with arrows. The peak value of the dry asphalt curve is marked by horizontal and vertical dashed lines, indicating peak adhesion and optimal slip ratio, respectively. The current working point is marked on the wet asphalt curve: the vertical dashed line passing through the working point and the circle mark the corresponding point value characteristics, with the definition of the point value ratio labeled with a leader line; the oblique line segments on both sides of the working point are local tangents, corresponding to local slope characteristics, with the definition of the slope ratio labeled with a leader line; the shaded area near the working point, defined by two thin vertical lines, is the integration interval for the cumulative characteristics of the slip domain, with the upper and lower limits of the interval marked, and the definition of the area ratio labeled with a leader line.

[0091] In some embodiments of this application, a family of load-related adhesion characteristic reference curves is pre-defined. Specifically, M typical road surface adhesion characteristic reference curves are established offline, and the tire normal load is explicitly introduced as a second-dimensional independent variable: , , In the formula, Number the reference curve; Number of reference curves; For the first The curve at slip ratio With normal load The adhesion coefficient at the point is dimensionless. For this curve under load The peak value below is dimensionless; The normalized shape function is dimensionless and satisfies... ; For this curve under load The optimal slip ratio under the given conditions is dimensionless; This represents the peak adhesion of the road surface under the reference load. The reference normal load for this axis is taken as the static axis load, and the unit is Newton; and The coefficients of the first and second terms of load sensitivity are dimensionless and obtained by fitting tire test bench data.

[0092] In the embodiments of this application, the normal load dimension is introduced based on the consideration that the tire has load sensitivity when the active suspension changes the normal load. If the adhesion characteristic reference curve is only a univariate function of the slip ratio, then every time the controller changes the normal load, the reference characteristic will cause a systematic deviation from the real tire, making it difficult to achieve closed-loop coupling between adhesion estimation and suspension control.

[0093] In some embodiments of this application, performing working condition matching and effective domain screening on a pre-defined set of adhesion characteristic reference curves refers to selecting a subset of adhesion characteristic reference curves from the set that match the current working state based on the normal load, slip ratio, and adhesion coefficient, and determining the effective slip ratio range of each adhesion characteristic reference curve in the subset. The purpose of this process is to ensure that the adhesion characteristic reference curves participating in subsequent feature extraction match the current working state in both the load and slip ratio dimensions, avoiding systematic deviations introduced due to load deviations or inappropriate slip ratio ranges.

[0094] In some embodiments of this application, the point value feature is the ratio of the adhesion coefficient of the current working state to the adhesion coefficient of the j-th adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the current slip ratio and normal load, that is: , In the formula, The ratio of point values ​​is dimensionless. The adhesion coefficient is used for the current working state; is the adhesion coefficient of the j-th adhesion characteristic reference curve at the current slip ratio and normal load.

[0095] In some embodiments of this application, the cumulative feature of the slip domain is the ratio of the cumulative area of ​​the measured adhesion characteristics to the cumulative area of ​​the j-th adhesion characteristic reference curve within the effective slip ratio range, i.e.: ; In the formula, This is a ratio of areas, dimensionless; This represents the cumulative area of ​​the measured adhesion characteristics within the effective slip ratio range. Let be the cumulative area of ​​the j-th adhesion characteristic reference curve within the same effective slip ratio range.

[0096] In some embodiments of this application, the local slope feature is the ratio of the measured local slope at the current working state to the slope of the j-th adhesion characteristic reference curve at the current working state, that is: , In the formula, The ratio of slopes is dimensionless. This represents the measured local slope at the current operating state. Let be the slope of the j-th adhesion characteristic reference curve in the current operating state.

[0097] In some embodiments of this application, point value features are obtained as follows: at the current slip ratio, the attachment and the first The ratio of the adhesion characteristic reference curve is taken and mapped to the peak value of the curve. , In the formula, The ratio of point values ​​is dimensionless. For the first The peak attachment candidate estimate obtained from the 3 curves is dimensionless. The above formula gives a candidate value for each curve, for a total of 3 curves. One. Likelihood weights are constructed using matched residuals to converge it into a single-channel estimate: , In the formula, To match the residuals, dimensionless; The point-value channel residual scale parameter is taken as a value of the same order of magnitude as the measured adhesion coefficient noise standard deviation; These are the likelihood weights, which are non-negative and sum to 1. For point-value channel peak attachment estimation; The candidate dispersion is used as a component of the uncertainty of this channel. This is to achieve a dummy summation. The purpose of this convergence step is to: when multiple attachment characteristic reference curves are difficult to distinguish at the current operating point, the candidate values ​​are significantly dispersed, and the dispersion automatically increases, thereby reducing the weight of this channel in subsequent fusion.

[0098] In practice, the point value channel is processed according to the following procedure: For each adhesion characteristic reference curve j = 1, 2, …, M in the reference curve family, a candidate value is obtained. ; Use matching residuals Construct likelihood weights Point value channel estimation can be obtained. And obtain the candidate dispersion. .

[0099] The cumulative channel, slope channel, and point value channel are processed using the same procedure as described above: the cumulative channel determines a candidate value for each attachment characteristic reference curve j. Then construct likelihood weights based on area residuals. Converging it into a single-channel estimate yields the cumulative channel estimate. And obtain the candidate dispersion. .

[0100] Slope channel satisfies and The set of curves Each adhesion characteristic reference curve j in the data determines a candidate value. Then construct the likelihood weights based on the slope residuals. Converging it into a single-channel estimate yields the slope channel estimate. And obtain the candidate dispersion. .

[0101] In some embodiments of this application, the cumulative features of the sliding domain are obtained by: [using a sliding window] Triples within Project to Each box has an equal width and a slip ratio, with a box width of [missing information]. , No. The center slip rate of each box is The sample size is Using the adhesion mean The mean value of the normal load is The integral interval and validity criterion are as follows: , In the formula, For sliding data windows; This represents the total number of boxes. This refers to the width of the container. For box numbering; and These are the lower and upper limits of the integration interval, respectively; This is the lower threshold for slip ratio effectiveness; The measured maximum slip ratio within the window; This is the upper limit of the allowable slip ratio, used to avoid the decline phase after reaching the peak; Minimum number of samples required for effective binning; Effective box fill rate; symbol Indicates the number of elements in the set; The set of valid bins for participating in the points calculation.

[0102] In some embodiments of this application, the conditions under which this channel is valid are as follows: and ,in Minimum effective interval width, This represents the minimum fill rate; if either of these conditions is not met, this channel is invalidated. The purpose of this adaptive interval rule is to ensure that the area ratio is only used when the measured data and the reference curve are taken from the same slip interval.

[0103] In some embodiments of this application, the cumulative area of ​​the measured adhesion characteristics and the cumulative area of ​​the j-th adhesion characteristic reference curve are respectively... and A candidate value is given for each adhesion characteristic reference curve. Likelihood weights are constructed using area residuals. Converging it into a single-channel estimate The cumulative characteristics of the slip domain are .

[0104] Specifically, the calculation formula is as follows: , In some embodiments of this application, window data is calculated according to a forgetting factor. Weighted, with values ​​between 0 and 1, and cleared when a sudden change in road surface adhesion is detected.

[0105] In some embodiments of this application, the local slope characteristics are obtained as follows: the slope is calculated using a locally weighted linear regression over the slip domain, i.e.: , In the formula, The regression intercept is the smoothing effect at the current working point using attachment. The measured local slope obtained from the regression; and The intercept and slope are to be optimized; The sample number within the window; The current time number; For the slip ratio domain kernel function; and The first The absolute value of the slip ratio of each sample and the application of adhesion; The slip ratio at the current operating point; the variance of the regression residuals is denoted as... .

[0106] In some embodiments of this application, a candidate value is given for each adhesion characteristic reference curve. And construct likelihood weights based on slope residuals. Converging it into a single-channel estimate ;Specifically, , In the formula, For the first The slope of the reference curve at the current working point; The ratio of slopes is dimensionless. Candidate estimates; The likelihood weights are constructed based on the slope residuals; For slope channel peak attachment estimation; The set of curves participating in this channel weighting only includes those that simultaneously satisfy... and The curve, in which The minimum effective value of the reference slope This is the peak exit coefficient, taken as a positive number less than 1; the purpose of this set is to prevent the slope ratio from diverging when the reference slope approaches zero. Candidate dispersion. The calculation method and same.

[0107] In some embodiments of this application, the working status information also includes longitudinal force; multiple attachment curve features are weighted and fused based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty, which is achieved through the following steps: Step B01: Map the point value features, slip domain cumulative features, and local slope features to the same peak attachment space to obtain the point value channel peak attachment candidate value, the cumulative channel peak attachment candidate value, and the slope channel peak attachment candidate value.

[0108] Step B02: For each channel, construct the total variance of the channel based on the sensitivity Jacobian matrix of the observation vector to the peak attachment candidate value of the channel, the observation noise covariance matrix, the variance of the matching residual between the channel and the attachment characteristic reference curve, and the candidate dispersion; the observation vector is a three-dimensional vector composed of slip ratio, longitudinal force and normal load.

[0109] Step B03: Construct inverse variance fusion weights based on the total variance of each channel, and use them as channel weights.

[0110] Step B04: Based on the channel weights, the peak attachment candidate values ​​of each channel are weighted and fused to obtain the peak attachment estimate.

[0111] Step B05: Based on the channel weights, the total variance of each channel, and the difference between the peak attachment candidate value and the peak attachment estimate of each channel, the estimation uncertainty is obtained.

[0112] The candidate dispersion is determined by the weighted dispersion of the candidate values ​​obtained from the reference curves of each adhesion characteristic within the same channel relative to the channel estimate of that channel.

[0113] In the embodiments of this application, the physical dimensions of the multiple attachment curve features are different, making direct weighting physically impossible; moreover, the reliability of various features varies greatly in different slip intervals, and fixed weights cannot reflect this difference. Therefore, instead of weighting in the feature space, the three features are first mapped to the same peak attachment space, and then a variance is uniformly constructed in this space according to uncertainty propagation and inverse variance weighting is performed.

[0114] In some embodiments of this application, the method includes a confidence construction operation based on the sensitivity Jacobian matrix; the confidence construction operation based on the sensitivity Jacobian matrix includes: taking the observation vector and its noise covariance matrix as... , In the formula, This is a three-dimensional observation vector; To observe the noise covariance matrix; , , These are the estimated noise standard deviations for slip ratio, longitudinal force, and normal load, respectively. The first... Channel estimation is written as a mapping of observations. ,in Choose one of P, I, and D, and define: , , In the formula, For the first The channel is a one-row, three-column sensitivity Jacobian matrix from observations to peak attachment estimation; This represents the total variance of the channel. The variance of the matching residuals between this channel and the reference curve is taken for the slope channel. ; This represents the candidate dispersion. This process automatically internalizes the discriminability of the adhesion characteristic reference curve into variance.

[0115] In some embodiments of this application, the first... Separability index of class features , In the formula, It is the separability index; For the first The first reference curve Class feature value, taking point value channel , take the cumulative channel Take the slope channel ; and Number the curves. When different road surfaces have very similar characteristics at the current working point, that is... When the value is very small, the sensitivity of peak attachment deduced from the features must be very high, that is... and Increasing to the same order, thus Automatically amplifies and automatically reduces the channel weight, eliminating the need for manual adjustment of fixed weights.

[0116] In some embodiments of this application, the point value channel, cumulative channel, and slope channel are each equipped with a gating function; inverse variance fusion weights are constructed based on the total variance of each channel as channel weights, including: The inverse variance of each channel is obtained by performing the inverse operation on the total variance of each channel. The channel weights are obtained by considering the inverse variance of each channel and the corresponding gating function value.

[0117] For example, , In the formula, , , These are the gate function values ​​for the point value, cumulative value, and slope channel, respectively. This is an indicator function, taking the value 1 if the condition is true and 0 otherwise; This is the longitudinal force excitation threshold, measured in Newtons (N). symbol This indicates a logical AND operation. The optimal slip ratio is the current maximum likelihood reference curve.

[0118] In some embodiments of this application, the fusion weight and fusion estimate are: , In the formula, For the first The channel fusion weight takes a non-negative value, and the sum of the fusion weights of each channel is 1. For peak attachment estimation after fusion; To achieve a dummy summation, when the gating function of all channels is zero, the peak attachment estimate is not updated, but the prior estimate of the peak attachment from the previous cycle is maintained, and the estimation uncertainty of the prior estimate of the peak attachment from the previous cycle is increased.

[0119] In some embodiments of this application, the process of determining the peak attachment estimate and the estimation uncertainty further includes: correcting the inverse variance fusion result through the channel correlation matrix to obtain the corrected estimation uncertainty; and applying a lower bound clamp to the corrected estimation uncertainty so that the estimation uncertainty is not lower than the set proportion of the optimal channel uncertainty participating in the fusion.

[0120] In practice, the point value features, slip domain cumulative features, and local slope features extracted from the three channels are all derived from the same set of observation vectors, and there is a strong correlation between their errors. If they are processed independently, the fusion variance will be systematically underestimated, leading to an overestimation of the conservative adhesion parameter and a risk of overestimating the adhesion capability in the safety direction. Therefore, a channel correlation matrix is ​​introduced for correction, and a lower bound clamp is applied. , In the formula, This is the 3D fusion weight vector; It is a third-order channel covariance matrix; For the first Channel and the The error correlation coefficient between channels, with values ​​between -1 and 1, is calibrated by offline Monte Carlo simulation and stored as a lookup table; The first term on the right-hand side of the third equation is the weighted covariance term, and the second term is the discrete term caused by the estimation difference between channels. The arrows in the fourth equation indicate substitution assignment. This is the uncertainty lower bound clamping coefficient, with a value between 0 and 1. Its meaning is that the uncertainty of the fusion result must not be lower than the uncertainty of the optimal channel participating in the fusion. times.

[0121] Candidate dispersion, taking the point value channel as an example, is obtained as follows: The matching residual is calculated for each adhesion characteristic reference curve. And construct likelihood weights from the matching residuals. In the formula, The scaling parameter for the point-value channel residuals is taken as a value of the same order of magnitude as the standard deviation of the measured adhesion coefficient noise; the channel estimate for the point-value channels is obtained by the likelihood weights. Candidate dispersion is The calculation method for candidate dispersion in the cumulative channel and slope channel is the same as that in the point value channel.

[0122] In some embodiments of this application, the conservative adhesion parameter is obtained by subtracting a set multiple of the estimation uncertainty from the peak adhesion estimate, and a clamping condition is applied to the conservative adhesion parameter by the lower and upper bounds of the adhesion coefficient.

[0123] In practical implementation, the conservative attachment parameters can be determined by the following formula: , In the formula, For the first Conservative adhesion parameters of the shaft; and These represent the peak attachment estimation and fusion uncertainty of the axis, respectively. This is a conservative coefficient used only to compensate for estimation uncertainty; its value corresponds to the required one-sided confidence level. and These are the lower and upper bounds of the adhesion coefficient, respectively, taking typical boundary values ​​for ice surfaces and dry asphalt pavements; To be Limited to the range Clamping functions within.

[0124] In some embodiments of this application, when all channel gating is zero, i.e., the excitation is insufficient to support the update, instead of forcibly outputting an accurate estimate, the prior estimate of the peak value from the previous control cycle is maintained and the variance is expanded: , In the formula, The uncertainty expansion rate is expressed in seconds. The estimator update cycle is in seconds. This process enables the controller to distinguish between two states: high adhesion and an inability to determine the state due to insufficient excitation.

[0125] In some embodiments of this application, the method further includes: pavement adhesion mutation detection; pavement adhesion mutation detection and inter-axle feedforward, including: normalizing the information residual. As a mutation criterion; when continuous Each cycle satisfies Upon detecting an attachment mutation, the slip domain data buffer is immediately cleared, the cumulative feature channel gating is set to zero, and the fusion uncertainty is set to the prior upper limit. The number of consecutive periods required for mutation determination This is the threshold for mutation criteria.

[0126] Normalized new information residual It can be determined by the following formula: , In the formula, The reference curve number for the adhesion characteristic with the highest current likelihood weight; The modeling variance of the adhesion characteristic reference curve itself.

[0127] In some embodiments of this application, the method further includes: inter-axle feedforward; inter-axle feedforward includes: for rear-wheel drive vehicles, further utilizing inter-axle time delay to construct a front axle-to-rear axle attachment feedforward prediction:

[0128] In the formula, The rear axle feedforward prediction value is obtained from the front axle estimated delay; For continuous time; The time difference between the front and rear wheels passing through the same point on the road surface, in seconds; For rear axle secondary fusion estimation; The uncertainty of the feedforward forecast depends on the vehicle speed estimation error and the lateral consistency of the road surface. This mechanism aims to shorten the reconvergence time of the drive axle after abrupt changes in road surface adhesion.

[0129] In some embodiments of this application, based on conservative attachment parameters and prediction parameter information, the minimum normal load requirement information and the vehicle's tire normal load prediction information are obtained step by step. The attachment feasible region constraint is then obtained by comparing the minimum normal load requirement information and the normal load prediction information. This can be achieved through the following steps: Step C01: Receive the longitudinal force demand sequence in the predicted time domain given by the longitudinal control system, and allocate the longitudinal force demand sequence to the front axle and rear axle according to the drive arrangement to obtain the longitudinal force demand of the target axle.

[0130] Step C02: Based on the conservative adhesion parameters, maximum adhesion utilization rate, and target axis longitudinal force requirement, obtain the minimum normal load requirement information for each prediction step.

[0131] Step C03 involves predicting the normal load prediction information on the same time axis using the half-vehicle model and the forward road elevation prediction sequence, and comparing the normal load prediction information with the minimum normal load requirement information step by step.

[0132] Step C04: When the predicted normal load information is not less than the minimum normal load requirement information, the prediction step is determined to be feasible; otherwise, the prediction step is determined to be infeasible. The feasibility determination of each prediction step constitutes the attached feasible domain constraint.

[0133] In some embodiments of this application, the method includes: road surface elevation prediction and inter-axle time delay alignment; the road surface elevation prediction and inter-axle time delay alignment specifically includes: , In the formula, The prediction step number, with values ​​ranging from 1 to... ; To predict the number of time-domain steps; The discrete step number corresponding to the inter-axis time delay; symbol This indicates rounding up. The purpose of the above formula is to show that, with the same forward-looking sensor configuration, the effective pre-aiming time obtained by the rear axle is longer than that of the front axle. For rear-wheel drive vehicles, the adhesion bottleneck in driving conditions is located at the rear axle. This characteristic gives the drive axle, which is the bottleneck, the longest anti-collision margin.

[0134] In some embodiments of this application, the inter-axle distribution of longitudinal forces in a rear-wheel-drive vehicle is as follows: , In the formula, For the first The predicted longitudinal force requirement of the entire vehicle is expressed in Newtons (N), with positive for driving and negative for braking. and These are the longitudinal force requirements allocated to the front and rear axles, respectively. The front axle is assigned a braking force distribution ratio, with a value between 0 and 1. Under driving conditions, all demand is concentrated on the rear axle, making it a bottleneck for adhesion; under braking conditions, the rear axle dynamically reduces load due to load transfer, which also poses a risk of insufficient adhesion margin.

[0135] In some embodiments of this application, the method further includes: determining the minimum normal load requirement and feasible region step by step, as follows: , In the formula, For the first Axis at the Predict the attachment utilization rate of the step; To determine the maximum allowable adhesion utilization, a safety threshold less than 1 is selected. The minimum normal load, measured in Newtons, is required to meet the longitudinal force demand. Each prediction step compares the predicted normal load with this minimum requirement, thus determining the adhesion feasibility region: a prediction step is feasible if the predicted normal load is not less than the minimum requirement; otherwise, it is not. This constraint differs fundamentally from traditional tire-on-ground constraints, which ensure a positive normal load while ensuring sufficient adhesion margin. Conservative coefficient. With utilization limit The division of labor is clearly distinguished: the former compensates for the uncertainty of adhesion estimation, while the latter is used to compensate for tire model errors and execution errors. The two are calibrated separately to avoid the repeated superposition of conservative quantities, which would make the constraints frequently infeasible on low-adhesion road surfaces.

[0136] In some embodiments of this application, the method further includes: extrapolating the attenuation of the conservative attachment parameter in the prediction time domain; the extrapolation of the attenuation of the conservative attachment parameter in the prediction time domain is implemented based on the following formula: , The meanings of the symbols in the formula are the same as those in the previous text. The purpose of this formula is to ensure that the further the prediction step, the greater the uncertainty and the lower the conservative attachment parameter, so as to avoid extrapolating the estimate at the current moment to the entire prediction time domain without attenuation.

[0137] S103, under constraints of working force amplitude, working force change rate, suspension travel, tire contact with the ground, and adhesion feasible region, construct a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and working force energy consumption as the objective function, and determine the front and rear axle working forces when the objective function satisfies the constraints.

[0138] In some embodiments of this application, determining the front and rear axis driving forces when the objective function satisfies the constraints includes: performing a quadratic programming solution operation on the objective function and constraints based on the Laguerre basis function matrix to obtain the front and rear axis driving forces when the objective function satisfies the constraints.

[0139] In some embodiments of this application, the prediction equation is: , , In the formula, To achieve a dummy target; It consists of a road surface elevation prediction sequence and a longitudinal force demand sequence.

[0140] The controlled output is:

[0141] In the formula, The vertical acceleration on the spring is expressed in meters per second squared, which characterizes the comfort of riding in the floating tunnel. Pitch angular acceleration, measured in radians per second squared, characterizes the comfort of the pitch tunnel ride. and These are the front and rear suspension dynamic deflections, respectively, in meters; and It is a four-dimensional unit coordinate vector, with its first and second components being 1 and the rest being 0.

[0142] In some embodiments of this application, the objective function is: , In the formula, The objective function value; to The controlled output weight coefficients are as follows: and These correspond to the comfort trade-offs of the buoyancy and pitch channels, respectively. and Corresponding to the dynamic deflection of the front and rear suspensions. and Corresponding dynamic loads on the front and rear tires; and This represents the increment of the tire's dynamic load relative to the static axle load. and As a weighting coefficient for power energy consumption; For attachment constraint slack variables, in units of Newtons, non-negative; and These are the first and second penalty weights for the slack variables, respectively. The first penalty term is used to ensure that the slack variables are exactly zero when a feasible solution exists.

[0143] In some embodiments of this application, the constraint set is as follows: , In the formula, For the first Maximum operating force of the shaft actuator; The maximum allowable change in working power within a control cycle; Maximum permissible suspension dynamic deflection; The minimum normal load required to keep the tires from leaving the ground; To maintain the set upper limit of the normal load for excitation, it only takes effect when the condition of suppressing the normal load is met. The first two equations are actuator amplitude and rate of change constraints, the third equation is suspension travel constraint, the fourth equation is tire contact constraint, the fifth and sixth equations are soft constraints of the adhesion feasible region, and the seventh equation is excitation holding constraint.

[0144] In some embodiments of this application, the method includes: Laguerre dimensionality reduction parameterization; the Laguerre dimensionality reduction parameterization is: , In the formula, For the first Step the Laguerre basis function matrix; For the first shaft Wikipedia function vectors; Let Laguerre coefficients be the vector to be determined; This indicates a diagonal arrangement of blocks. The basis functions are generated by the following recursion and initial values: , In the formula, For the first The poles of the Laguerre network are axial, with values ​​between 0 and 1, and their magnitudes determine the decay rate of the basis functions. Let Laguerre order be the number of the axis. for The lower triangular state transition matrix is ​​denoted as follows: Its elements are: , In the formula, and These are the row and column numbers of the matrix, respectively. Substituting them, the original problem is reduced to a small-scale quadratic programming problem: , , In the formula, For decision variable vectors; The Hessian matrix; This is the vector of coefficients for the first-order terms; and These are the constraint matrix and constraint vector obtained by rearranging the inequality constraints, respectively. The free response matrix is ​​derived from... It is formed by stacking powers of each other; The forced response matrix is ​​formed by... , It is formed by convolving with the basis function matrix; The disturbance response matrix is ​​given by... and Convolutional; To predict the stacked vector of the perturbation sequence in the time domain; To output a weighted diagonal matrix; The input is the relaxation weighting matrix.

[0145] In some embodiments of this application, after using Laguerre dimensionality reduction parameterization, the dimension of the decision variables is compressed from the total number of driving forces in the prediction time domain to a few coefficients, typically of order three to six, which is sufficient to adequately approximate the original control sequence. The embodiments of this application use Laguerre dimensionality reduction parameterization, retaining the dimensionality reduction effect of Laguerre, but do not employ the solution method of obtaining unconstrained analytic stationary points by differentiating the coefficients. The attached feasible region is the core safety constraint, which is addressed through explicit inequalities. After obtaining the optimal coefficients, only the driving force at the current time step is executed, and the solution is recalculated in the next control cycle. The driving force at the current time step can be determined by the following formula: .

[0146] The above embodiments of this application physically accomplish two types of redistribution: time-dimensional redistribution, that is, shifting the trough of normal load from the period of high longitudinal force demand to the period of low demand within the prediction time domain; and inter-axle redistribution, that is, using the pitch channel as the dynamic component to transfer transient load between the front and rear axles, further effectively improving the accuracy of suspension control and enhancing the stability of vehicle driving.

[0147] S104 uses the front and rear axles as power sources to implement suspension control for the vehicle.

[0148] In some embodiments of this application, such as Figure 5 As shown, based on the front and rear axles as power sources, suspension control of the vehicle is implemented through the following steps: S501, determine whether the attached feasible region constraint is feasible.

[0149] S502, if the result of the judgment is that it is feasible, then the suspension control of the vehicle is implemented according to the front and rear axles as the driving force.

[0150] S503, if the judgment result is unachievable, then the vehicle suspension control is implemented according to the target front and rear axle working forces. The target front and rear axle working forces are obtained through the following process: under working force amplitude constraints, working force change rate constraints, and suspension travel constraints, the achievable normal load envelope of each axle in the prediction time domain is determined; from the achievable normal load envelope, conservative adhesion parameters, and maximum adhesion utilization rate, the maximum allowable longitudinal tire force of each axle is obtained; the maximum allowable longitudinal tire force is converted into an upper limit value of the control force parameter, and the upper limit value of the control force parameter is sent to the vehicle's longitudinal control system; and by softening the adhesion feasible domain constraints, the front and rear axle working forces are re-determined, and the re-determined front and rear axle working forces are used as the target front and rear axle working forces; the upper limit value of the control force parameter includes the upper limit of the driving torque and the upper limit of the braking force.

[0151] In some embodiments of this application, the envelope under the reachable normal load is determined by the following formula: , , In the formula, For the first Envelope under axial reachable normal load, in Newtons; The lower bound variable is used as an auxiliary variable, with the unit being cows; and This consists of constraint matrices and vectors containing only the dynamic amplitude, rate of change, and suspension travel constraints, excluding attachment constraints; (symbols omitted) This indicates that the condition holds true for all prediction steps. The minima problem in the first equation is transformed into a linear programming problem in the second equation by introducing auxiliary variables.

[0152] In some embodiments of this application, the method includes: determining the allowable longitudinal tire force; the allowable longitudinal tire force can be obtained by the following formula: , In the formula, For the first The maximum allowable longitudinal tire force on the axle, in Newtons; This represents the upper limit of the permissible drive wheel end torque under rear-wheel drive conditions, expressed in Newton-meters.

[0153] In some embodiments of this application, the upper limit of total braking force and the suggested braking force distribution ratio correction amount are simultaneously fed back during braking conditions.

[0154] In the above embodiment, the longitudinal domain requests the required longitudinal force from the suspension domain, and the suspension domain responds to the longitudinal domain with the maximum longitudinal force that can be safely supported based on the future road surface and the current actuator capabilities, thus forming a two-way coordination. This upper limit is caused by the limited capabilities of the active suspension actuators, rather than by the road surface adhesion itself.

[0155] In some embodiments of this application, to avoid the formation of an algebraic loop between the feedback and the longitudinal force requirement, an arbitration rule that does not iterate within a single period is specified, including: the first The periodic attachment constraint adopts the first The longitudinal force demand after periodic arbitration is directly issued after the upper limit calculated in this period is saturated and recalculated without being substituted back, thus ensuring that the amount of calculation and execution time of each control period are determined.

[0156] In some embodiments of this application, the denominator of the attachment is the normal load being adjusted by the controller, thus the attachment estimation and suspension control form a closed loop. When the controller increases the normal load, it utilizes the decrease in attachment, the return of the operating state to the origin, the weakening of the identification excitation, the increase in estimation uncertainty, and the decrease in conservative attachment parameters to demand a higher normal load, forming a positive feedback degradation loop. The following three measures are taken to suppress the normal load: First, dual time scale separation is used to determine the estimator update cycle. and Not less than 5, of which The time scale separation factor makes the estimated dynamics significantly slower than the control dynamics; Secondly, during the rapid change of power, the update is frozen when any axis meets the following conditions. The current period's estimate update is paused, and variance is inflated only according to the prior preservation rule, where The threshold for dynamic change used to determine freezing is expressed in Newtons (N). Third, the incentive maintains the constraint, when a certain axis simultaneously satisfies and The upper bound constraint of the normal load is activated, and the minimum identifiable excitation is actively retained under the premise of sufficient safety margin; once the attachment utilization rate rises above the safety threshold, the constraint is immediately released. In the formula, The threshold for maintaining attachment utilization as an incentive; The uncertainty threshold is used to incentivize the maintenance. The detection criterion for closed-loop degradation is continuity. Within the estimated period Monotonically rising and Simultaneously decreasing, among which The number of consecutive estimated cycles required for degradation determination.

[0157] In some embodiments of this application, the method further includes: fail-safe and graded degradation operations; the fail-safe and graded degradation operations include: When wheel speed, longitudinal force, or suspension displacement signals fail, the conservative adhesion parameters are directly set to the lower bound, the controller degenerates into comfort optimization without adhesion constraints, and at the same time issues a conservative upper limit for longitudinal force to the longitudinal system. When there is no feasible solution for the quadratic programming problem, relax the constraints in the following order: first, remove the incentive-maintaining constraints; second, increase the weight of the slack variables until the attached constraints are completely softened; finally, degenerate into an unconstrained Laguerre analytical solution containing only dynamic and travel constraints. When the quadratic programming solution exceeds the single-cycle time budget, the shift value of the optimal sequence of the previous cycle is used as the output of the current cycle, and the timeout event is included in the supervision unit. The monitoring unit continuously records the activation duration of the relaxation variable, the number of times the attachment constraint goes out of bounds, and the number of times the longitudinal force is allowed to be triggered. When the threshold is exceeded, the attachment safety function degradation status is reported to the vehicle network.

[0158] The suspension control method based on the adhesion feasible region constraint described in the above embodiments of this application includes: monitoring the vehicle's operating state to obtain the vehicle's working state information; determining the adhesion feasible region constraint based on the working state information and predicted parameter information; constructing a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and adhesion feasible region constraint as the objective function, and determining the front and rear axle actions when the objective function satisfies the constraint conditions; and implementing suspension control on the vehicle based on the front and rear axle actions. This method improves the accuracy of suspension control and enhances vehicle driving stability by converting adhesion capability into an adhesion feasible region constraint and introducing it into model predictive control.

[0159] Based on the same inventive concept, embodiments of this application also provide a suspension control device based on attachment feasible region constraints. For example... Figure 6 As shown, the device includes: a state information acquisition unit 601, an attachment constraint determination unit 602, a control parameter solving unit 603, and a suspension control execution unit 604. Wherein: The status information acquisition unit 601 is used to monitor the operating status of the vehicle and obtain the vehicle's working status information; the working status information includes the normal load of the vehicle's tires, slip ratio, and coefficient of adhesion. The adhesion constraint determination unit 602 is used to determine the adhesion feasible domain constraint based on the working status information and the prediction parameter information; the prediction parameter information includes the maximum adhesion utilization rate and the longitudinal force demand sequence and the forward road surface elevation prediction sequence in the prediction time domain. The control parameter solving unit 603 is used to construct a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load and working power energy consumption as the objective function under the constraints of working power amplitude, working power change rate, suspension travel, tire contact constraint and adhesion feasible region constraint, and to determine the front and rear axle working power when the objective function satisfies the constraint conditions. The suspension control actuator 604 is used to implement suspension control of the vehicle based on the power applied to the front and rear axles.

[0160] In one possible implementation, the attachment constraint determination unit 602 is specifically used for: Based on the normal load, slip ratio, and adhesion coefficient, feature extraction is performed on a pre-defined set of adhesion characteristic reference curves to obtain multiple adhesion curve features. Multiple attachment curve features are weighted and fused based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty. Based on the conservative adhesion parameters and prediction parameters, the minimum normal load requirement information and the vehicle tire normal load prediction information are obtained step by step. The adhesion feasible region constraint is obtained by comparing the minimum normal load requirement information and the normal load prediction information. The conservative adhesion parameters are obtained from the peak adhesion estimate and the estimation uncertainty.

[0161] In one possible implementation, the attachment constraint determination unit 602 is specifically used for: Based on the normal load, slip ratio, and adhesion coefficient, the pre-set set of adhesion characteristic reference curves is matched with the working conditions and the effective domain is screened to obtain a subset of adhesion characteristic reference curves that match the current working state, and the effective slip ratio range of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves is determined. Based on the subset of adhesion characteristic reference curves and the effective slip ratio range, the adhesion coefficient of the current working state is compared with the adhesion coefficient of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the same normal load and the same slip ratio to obtain multiple adhesion curve features; the multiple adhesion curve features include at least two of the following: point value features, slip domain cumulative features, and local slope features.

[0162] In one possible implementation, the working status information also includes longitudinal force; the attachment constraint determination unit 602 is specifically used for: Point value features, slip domain cumulative features, and local slope features are mapped to the same peak attachment space to obtain peak attachment candidate values ​​for point value channels, cumulative channel peak attachment candidate values, and slope channel peak attachment candidate values. For each channel, the total variance of the channel is constructed based on the sensitivity Jacobian matrix of the observation vector to the peak attachment candidate value of the channel, the observation noise covariance matrix, the variance of the matching residual between the channel and the attachment characteristic reference curve, and the candidate dispersion; the observation vector is a three-dimensional vector composed of slip ratio, longitudinal force and normal load. Construct inverse variance fusion weights based on the total variance of each channel, and use them as channel weights; The peak attachment candidate values ​​of each channel are weighted and fused based on the channel weights to obtain the peak attachment estimate; The estimation uncertainty is obtained based on the channel weights, the total variance of each channel, and the difference between the peak attachment candidate value and the peak attachment estimate of each channel. The candidate dispersion is determined by the weighted dispersion of the candidate values ​​obtained from the reference curves of each adhesion characteristic within the same channel relative to the channel estimate of that channel.

[0163] In one possible implementation, the point value channel, cumulative channel, and slope channel are each equipped with a gating function; the attachment constraint determination unit 602 is specifically used for: The inverse variance of each channel is obtained by performing the inverse operation on the total variance of each channel. The channel weights are obtained by considering the inverse variance of each channel and the corresponding gating function value.

[0164] In one possible implementation, the attachment constraint determination unit 602 is specifically used for: Receive the longitudinal force demand sequence in the predicted time domain given by the longitudinal control system, and distribute the longitudinal force demand sequence to the front axle and rear axle according to the drive arrangement to obtain the longitudinal force demand of the target axle; Based on conservative adhesion parameters, maximum adhesion utilization rate, and target axis longitudinal force requirements, the minimum normal load requirement information for each prediction step is obtained. The normal load prediction information on the same time axis is predicted by the half-vehicle model and the forward road elevation prediction sequence, and the normal load prediction information is compared with the minimum normal load requirement information step by step. When the predicted normal load information is not less than the minimum normal load requirement information, the prediction step is deemed feasible; otherwise, the prediction step is deemed infeasible. The feasibility determination of each prediction step constitutes the attached feasible region constraint.

[0165] In one possible implementation, the control parameter solving unit 603 is specifically used for: Based on the Laguerre basis function matrix, a quadratic programming solution is performed on the objective function and constraints to obtain the front and rear axis driving forces when the objective function satisfies the constraints.

[0166] In one possible implementation, the suspension control execution unit 604 is specifically used for: Determine whether the attached feasible region constraint is feasible; If the result of the judgment is that it is feasible, then the suspension control of the vehicle is implemented by using the front and rear axles as the power sources; If the result of the judgment is that it is not possible, then the suspension control of the vehicle is implemented according to the power of the front and rear axles of the target. The target's front and rear axles act as the driving force, which is obtained through the following process: Under constraints of dynamic amplitude, dynamic change rate, and suspension travel, the achievable normal load envelope of each axle in the prediction time domain is determined. The maximum allowable longitudinal tire force for each axle is obtained from the achievable normal load envelope, conservative adhesion parameters, and maximum adhesion utilization rate. This maximum allowable longitudinal tire force is converted into an upper limit value for control force parameters, which is then sent to the vehicle's longitudinal control system. By softening the adhesion feasible domain constraint, the front and rear axle driving forces are redefined, and these redefined forces are used as the target front and rear axle driving forces. The upper limit values ​​for control force parameters include the upper limit of driving torque and the upper limit of braking force.

[0167] Based on the same technical concept, this application also provides an electronic device, referring to... Figure 7 As shown, the electronic device is used to implement the methods described in the above-described method embodiments, for example, implementing... Figure 1 The method shown allows the electronic device to include a memory 701, a processor 702, an input unit 703, and a display panel 704.

[0168] The memory 701 stores computer programs executed by the processor 702. The memory 701 may mainly include a program storage area and a data storage area. The program storage area may store the operating system, applications required for at least one function, etc.; the data storage area may store data created based on the use of the electronic device, etc. The processor 702 may be a central processing unit (CPU) or a digital processing unit, etc. The input unit 703 may be used to acquire user input instructions. The display panel 704 is used to display information input by the user or information provided to the user. In this embodiment, the display panel 704 is mainly used to display the display interfaces of various applications in the terminal device and the control entities displayed in each display interface. Optionally, the display panel 704 may be configured as a liquid crystal display (LCD) or an OLED (organic light-emitting diode), etc.

[0169] This application embodiment does not limit the specific connection medium between the memory 701, processor 702, input unit 703, and display panel 704 described above. This application embodiment... Figure 7 The memory 701, processor 702, input unit 703, and display panel 704 are connected via a bus 705, and the bus 705 is in... Figure 7 The connections between other components are indicated by thick lines and are for illustrative purposes only, not as limiting information. The 705 bus can be divided into address bus, data bus, control bus, etc. For ease of illustration, Figure 7The bus is represented by a single thick line, but this does not mean that there is only one bus or one type of bus.

[0170] Memory 701 may be volatile memory, such as random-access memory (RAM); memory 701 may also be non-volatile memory, such as read-only memory, flash memory, hard disk drive (HDD), or solid-state drive (SSD); or memory 701 may be any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but is not limited thereto. Memory 701 may be a combination of the above-mentioned memories.

[0171] Processor 702 is used to invoke a computer program stored in memory 701 to execute as implemented. Figure 1 The example shown.

[0172] This application also provides a computer-readable storage medium storing computer-executable instructions required to execute the processor, including a program required to execute the processor.

[0173] In some possible implementations, various aspects of the suspension control method based on attachment feasible region constraints provided in this application can also be implemented as a program product, including program code. When the program product is run on a terminal device, the program code causes the terminal device to execute the steps of the suspension control method based on attachment feasible region constraints according to the various exemplary embodiments of this application described above. For example, an electronic device can perform such implementations. Figure 1 The example shown.

[0174] The program product may employ any combination of one or more readable media. A readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof.

[0175] The air conditioning control program product for a vehicle according to embodiments of this application can employ a portable compact disc read-only memory (CD-ROM) and include program code, and can run on a computing device. However, the program product of this application is not limited thereto. In this document, the readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0176] A readable signal medium may include a data signal propagated in baseband or as part of a carrier wave, carrying readable program code. This propagated data signal may take many forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. A readable signal medium may also be any readable medium other than a readable storage medium, capable of sending, propagating, or transmitting a program for use by or in conjunction with an instruction execution system, apparatus, or device.

[0177] The program code contained on the readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wired, optical fiber, RF, or any suitable combination thereof.

[0178] Program code for performing the operations of this application can be written in any combination of one or more programming languages, including entity-oriented programming languages ​​such as Java and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can execute entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server. In cases involving remote computing devices, the remote computing device can be connected to the user's computing device via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computing device (e.g., via the Internet using an Internet service provider).

[0179] It should be noted that although several units or sub-units of the device have been mentioned in the detailed description above, this division is merely exemplary and not mandatory. In fact, according to embodiments of this application, the features and functions of two or more units described above can be embodied in one unit. Conversely, the features and functions of one unit described above can be further divided and embodied by multiple units.

[0180] Furthermore, although the operations of the method of this application are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in that specific order, or that all the operations shown must be performed to achieve the desired result. Additionally or alternatively, certain steps may be omitted, multiple steps may be combined into one step, and / or one step may be broken down into multiple steps.

[0181] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0182] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable file processing device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable file processing device, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0183] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable file processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0184] These computer program instructions can also be loaded onto a computer or other programmable document processing device to cause a series of operational steps to be performed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable device for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0185] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0186] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A suspension control method based on attachment feasible region constraints, characterized in that, The method includes: The vehicle's operating status is monitored to obtain the vehicle's working status information, which includes the vehicle's tire normal load, slip ratio, and coefficient of adhesion. Based on the working status information and prediction parameter information, the adhesion feasible domain constraint is determined; the prediction parameter information includes the maximum adhesion utilization rate, the longitudinal force demand sequence in the prediction time domain, and the forward road surface elevation prediction sequence. Under constraints of working power amplitude, working power rate of change, suspension travel, tire contact with the ground, and the adhesion feasible region, a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load, and working power energy consumption is constructed as the objective function. The working power of the front and rear axles when the objective function satisfies the constraints is then determined. The suspension control of the vehicle is implemented based on the power generated by the front and rear axles.

2. The method according to claim 1, characterized in that, The step of determining the attached feasible region constraint based on the working state information and prediction parameter information includes: Based on the normal load, the slip ratio, and the adhesion coefficient, feature extraction is performed in a pre-defined set of reference adhesion characteristic information to obtain multiple adhesion curve features. The multiple attachment curve features are weighted and fused based on their corresponding channel weights to obtain the peak attachment estimate and the estimation uncertainty. Based on the conservative adhesion parameters and the prediction parameter information, the minimum normal load requirement information and the normal load prediction information of the vehicle's tires are obtained step by step. The adhesion feasibility region constraint is obtained by comparing the minimum normal load requirement information and the normal load prediction information. The conservative adhesion parameters are obtained by the peak adhesion estimate and the estimation uncertainty.

3. The method according to claim 2, characterized in that, The process involves extracting features from a pre-defined set of reference adhesion characteristic information based on the normal load, the slip ratio, and the adhesion coefficient, to obtain multiple adhesion curve features, including: Based on the normal load, the slip ratio, and the adhesion coefficient, the pre-set set of adhesion characteristic reference curves is matched with the working conditions and the effective domain is filtered to obtain a subset of adhesion characteristic reference curves that match the current working state, and the effective slip ratio range of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves is determined. Based on the subset of adhesion characteristic reference curves and the effective slip ratio range, the adhesion coefficient of the current working state is compared with the adhesion coefficient of each adhesion characteristic reference curve in the subset of adhesion characteristic reference curves at the same normal load and the same slip ratio to obtain multiple adhesion curve features; the multiple adhesion curve features include at least two of the following: point value features, slip domain cumulative features, and local slope features.

4. The method according to claim 2, characterized in that, The working status information also includes longitudinal force; the weighted fusion operation of the multiple adhesion curve features based on corresponding channel weights to obtain the peak adhesion estimate and the estimation uncertainty includes: The point value features, the slip domain cumulative features, and the local slope features are mapped to the same peak attachment space to obtain the point value channel peak attachment candidate value, the cumulative channel peak attachment candidate value, and the slope channel peak attachment candidate value. For each channel, the total variance of the channel is constructed based on the sensitivity Jacobian matrix of the observation vector to the peak attachment candidate value of the channel, the observation noise covariance matrix, the variance of the matching residual between the channel and the attachment characteristic reference curve, and the candidate dispersion; the observation vector is a three-dimensional vector composed of the slip ratio, the longitudinal force, and the normal load. Inverse variance fusion weights are constructed based on the total variance of each channel, and these weights are used as the channel weights. Based on the channel weights, the peak attachment candidate values ​​of each channel are weighted and fused to obtain the peak attachment estimate. The estimation uncertainty is obtained based on the channel weights, the total variance of each channel, and the difference between the peak attachment candidate value and the peak attachment estimate of each channel. The candidate dispersion is determined by the weighted dispersion of the candidate values ​​obtained from each attachment characteristic reference curve within the same channel relative to the channel estimate of that channel.

5. The method according to claim 4, characterized in that, The point value channel, the cumulative channel, and the slope channel are each equipped with a gating function; the construction of inverse variance fusion weights based on the total variance of each channel, as the channel weights, includes: The inverse variance of each channel is obtained by performing the inverse operation on the total variance of each channel. The channel weights are obtained based on the inverse variance of each channel and the corresponding gating function value.

6. The method according to claim 2, characterized in that, The process involves obtaining step-by-step minimum normal load requirement information and tire normal load prediction information based on conservative adhesion parameters and the prediction parameter information, and then comparing the minimum normal load requirement information and the normal load prediction information to obtain adhesion feasibility region constraints, including: Receive the longitudinal force demand sequence in the predicted time domain given by the longitudinal control system, and allocate the longitudinal force demand sequence to the front axle and rear axle according to the drive arrangement to obtain the longitudinal force demand of the target axle; Based on the conservative adhesion parameters, the maximum adhesion utilization rate, and the target axis longitudinal force requirement, the minimum normal load requirement information for each prediction step is obtained; The normal load prediction information on the same time axis is predicted by the half-vehicle model and the forward road elevation prediction sequence, and the normal load prediction information is compared with the minimum normal load requirement information step by step. When the predicted normal load information is not less than the minimum normal load requirement information, the prediction step is determined to be feasible; otherwise, the prediction step is determined to be infeasible. The feasibility determination of each prediction step constitutes the attached feasible domain constraint.

7. The method according to claim 1, characterized in that, Determining the front and rear axle forces when the objective function satisfies the constraints includes: Based on the Laguerre basis function matrix, a quadratic programming solution is performed on the objective function and constraints to obtain the front and rear axis driving forces when the objective function satisfies the constraints.

8. The method according to claim 2, characterized in that, The method of implementing suspension control of the vehicle based on the power generated by the front and rear axles includes: Determine whether the attachment feasible region constraint is feasible; If the result of the judgment is feasible, then the suspension control of the vehicle is implemented according to the power of the front and rear axles; If the result of the judgment is that it is not possible, then the suspension control of the vehicle is implemented according to the target front and rear axles as the driving force; The target's front and rear axles are powered by the following process: Under the constraints of the working force amplitude, the working force change rate, and the suspension travel, the achievable normal load envelope of each axle in the prediction time domain is determined. The maximum allowable longitudinal tire force of each axle is obtained from the achievable normal load envelope, the conservative adhesion parameters, and the maximum adhesion utilization rate. The maximum allowable longitudinal tire force is converted into an upper limit value for the control force parameters, and the upper limit value for the control force parameters is sent to the vehicle's longitudinal control system. By softening the adhesion feasible domain constraint, the front and rear axle working forces are re-determined, and the re-determined front and rear axle working forces are used as the target front and rear axle working forces. The upper limit value for the control force parameters includes an upper limit for the driving torque and an upper limit for the braking force.

9. A suspension control device based on attachment feasible region constraints, characterized in that, The device includes: A status information acquisition unit is used to monitor the operating status of the vehicle and obtain the working status information of the vehicle; the working status information includes the normal load, slip ratio and coefficient of adhesion of the vehicle's tires; The adhesion constraint determination unit is used to determine the adhesion feasible domain constraint based on the working status information and the prediction parameter information; the prediction parameter information includes the maximum adhesion utilization rate, the longitudinal force demand sequence in the prediction time domain, and the forward road surface elevation prediction sequence. The control parameter solving unit is used to construct a weighted quadratic form of sprung vertical acceleration, pitch angle acceleration, suspension dynamic deflection, tire dynamic load and working power energy consumption as the objective function under the constraints of working power amplitude, working power change rate, suspension travel, tire contact constraint and the adhesion feasible region constraint, and to determine the front and rear axle working power when the objective function satisfies the constraint conditions. A suspension control actuator is used to implement suspension control on the vehicle based on the power applied to the front and rear axles.

10. An electronic device, characterized in that, It includes a processor and a memory, wherein the memory stores program code that, when executed by the processor, causes the processor to perform the steps of the method according to any one of claims 1 to 8.

Images