CN111216732B

CN111216732B - Road surface friction coefficient estimation method and device and vehicle

Info

Publication number: CN111216732B
Application number: CN201811418159.XA
Authority: CN
Inventors: 约瑟夫.艾哈迈德.古奈姆; 孙玉; 牛小锋; 徐波; 张英富; 王彬彬; 孔凡茂; 陈建宏; 戴彦收; 郭君昌; 王海龙
Original assignee: Great Wall Motor Co Ltd
Current assignee: Great Wall Motor Co Ltd
Priority date: 2018-11-26
Filing date: 2018-11-26
Publication date: 2021-04-09
Anticipated expiration: 2038-11-26
Also published as: CN111216732A

Abstract

The invention relates to the technical field of vehicles, and provides a method and a device for estimating a road surface friction coefficient and a vehicle. The estimation method comprises the following steps: determining a plurality of sets of classifiers for calculating the road surface friction coefficient, wherein the sets of classifiers are determined from vehicle operation data and vehicle lateral control related data, and the vehicle lateral control related data comprises a tire slip angle of the vehicle, a tire lateral force, a first self-aligning torque (SAT) estimated based on the EPS system, a second SAT estimated based on the tire lateral dynamics model, and derivatives of mean and variance of the first SAT; selecting a corresponding classifier set to calculate a road surface friction coefficient according to a tire slip angle, a tire lateral force and a steering rate generated by a driver performing lateral operation on a vehicle; and determining the calculated maximum road friction coefficient as a final road friction coefficient. The invention utilizes the multi-set classifier, and can accurately detect the maximum friction coefficient of the road surface when the driver carries out lateral operation in different degrees.

Description

Road surface friction coefficient estimation method and device and vehicle

Technical Field

The invention relates to the technical field of vehicles, in particular to a method and a device for estimating a road surface friction coefficient and a vehicle.

Background

In driving a vehicle, the driver may encounter different road surfaces, such as asphalt, gravel, dry, wet, ice, snow, etc. These and other types of road surfaces each have different road surface coefficients of friction, and these different road surface coefficients of friction affect tire grip and vehicle stability.

For safety reasons, it is vital that the vehicle is capable of operating in a manner that allows it to respond quickly to various road conditions at any time, particularly when the vehicle is moving sideways, for example during cornering, lane changes or during obstacle avoidance. One way to solve this problem is to use an estimate of the instantaneous road friction. Therefore, it is desirable to know the road surface friction coefficient during vehicle operation. For example, a control system of a vehicle may use information about the road surface friction coefficient to control one or more vehicle components to assist a driver in operating the vehicle in a safe manner.

However, there is currently no method by which the road surface friction coefficient can be directly measured, and only and necessarily estimated using available sensor information on the vehicle. However, such methods for estimating the road surface friction coefficient are unreliable because they are sensitive to different vehicle dynamic behaviors, such as steering behavior, in that the estimated friction coefficient may vary greatly even on the same type of road surface, for example, on dry road surfaces, affected by different vehicle dynamic behaviors, such as steering behavior, and the road surface friction coefficient estimated using available sensors on the vehicle is likely to be a number of inequalities between 0.4 and 1. Therefore, this method is poor in reliability and robustness.

In addition, prior art methods of estimating road surface coefficient of friction require greater drive excitation (steering) to produce large tire slip angles before the maximum lateral coefficient of friction of the road surface at both small and large tire slip angles can be estimated. In practice, the driver needs to take a violent lateral operation to generate a large driving excitation, and if the operation trend of the driver is mild, the estimated friction coefficient is distorted, which obviously brings higher driving requirements to the driver and influences the driving experience of the driver.

Disclosure of Invention

In view of this, the present invention aims to provide a road surface friction coefficient estimation method with higher reliability and robustness, so as to solve the technical problem that the road surface friction coefficient can be measured only by a driver adopting a violent lateral operation in the prior art.

In order to achieve the purpose, the technical scheme of the invention is realized as follows:

a method of estimating a road surface friction coefficient, the method comprising: determining a plurality of sets of classifiers for calculating a road surface friction coefficient, wherein the sets of classifiers are determined based on vehicle operating data and vehicle lateral control related data, and the vehicle operation data comprises vehicle speed, lateral acceleration and steering rate of the vehicle, the vehicle lateral control related data comprises tire slip angle of the vehicle, tire lateral force, first self-aligning torque (SAT) estimated based on an electric power steering system, second SAT estimated based on a tire lateral dynamics model, and derivatives of mean and variance of the first SAT over a sampling time, wherein each classifier set comprises parameters and calculation models for calculating the road surface friction coefficient different from other classifier sets, wherein the parameter is part of data comprised by the vehicle operation data and/or data comprised by the vehicle lateral control related data; selecting a corresponding set of classifiers to calculate road surface friction coefficients as a function of the tire slip angle, the tire lateral force, and the steering rate resulting from a driver's lateral operation of a vehicle, wherein each set of classifiers is preconfigured to match a set of selection conditions comprising the tire slip angle, the tire lateral force, and the steering rate; and determining the calculated maximum road friction coefficient as a final road friction coefficient.

Compared with the prior art, the road surface friction coefficient estimation method provided by the invention has the advantages that the maximum road surface friction coefficient can be accurately detected by using the multi-set classifier when the driver performs lateral operations of different degrees, so that better vehicle lateral motion control is realized.

The present invention also provides a road surface friction coefficient estimation device, including: a classifier determination module for determining a plurality of sets of classifiers for calculating a road surface friction coefficient, wherein the set of classifiers is determined based on vehicle operation data and vehicle lateral control related data, and the vehicle operation data comprises vehicle speed, lateral acceleration and steering rate of the vehicle, the vehicle lateral control related data comprises tire slip angle of the vehicle, tire lateral force, first self-aligning torque (SAT) estimated based on an electric power steering system, second SAT estimated based on a tire lateral dynamics model, and derivatives of mean and variance of the first SAT over a sampling time, wherein each classifier set comprises parameters and calculation models for calculating the road surface friction coefficient different from other classifier sets, wherein the parameter is part of data comprised by the vehicle operation data and/or data comprised by the vehicle lateral control related data; a classifier selection module for selecting a corresponding set of classifiers to calculate road surface friction coefficients according to the tire slip angle, the tire lateral force and the steering rate resulting from a driver's lateral operation of a vehicle, wherein each set of classifiers is preconfigured to match a set of selection conditions comprising the tire slip angle, the tire lateral force and the steering rate; and the lateral friction coefficient determining module is used for determining the calculated maximum road friction coefficient as a final road friction coefficient.

The present disclosure also provides a machine-readable storage medium having instructions stored thereon for causing a machine to perform the above-described method of estimating a road surface friction coefficient.

The present invention also provides a vehicle comprising: a steering wheel configured to rotate in response to a driver steering input, wherein the steering input includes a steering torque and a steering angle; a torque sensor configured to measure a steering torque; a steering angle sensor configured to measure a steering angle; the steering mechanism realizes follow-up with the steering wheel through a steering column; the machine-readable storage medium described above; a controller configured to receive vehicle operation data including the steering torque and the steering angle and execute instructions stored in the machine-readable storage medium based on the vehicle operation data.

The estimation device, the machine-readable storage medium and the vehicle have the same technical problems and advantages as the estimation device in the prior art, and thus, the detailed description thereof is omitted.

Additional features and advantages of the invention will be set forth in the detailed description which follows.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate an embodiment of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:

FIG. 1 is a schematic diagram of the modules of a method for estimating road surface friction coefficient according to an embodiment of the present invention;

FIG. 2 is a schematic illustration of a vehicle having an EPS system and a controller for determining a road surface coefficient of friction in an embodiment of the present invention;

FIG. 3 is a schematic flow chart illustrating the determination of the occurrence of straight-ahead travel of a vehicle in an embodiment of the present invention;

FIG. 4 is a flowchart illustrating estimating a first SAT according to an embodiment of the invention;

FIG. 5 is a flow chart of calculating a derivative of the mean and variance of the first SAT in an embodiment of the invention;

FIG. 6 is a schematic flow chart of the determination of the linear region of lateral force dynamics in an embodiment of the present invention;

FIG. 7 is a schematic representation of an exemplary tire lateral dynamics model in an embodiment of the present invention;

FIG. 8 is a schematic representation of tire lateral force versus SAT (relative to tire slip angle;

FIG. 9 is a schematic free body diagram of a bicycle model in a turning maneuver in an embodiment of the present invention;

FIG. 10 is a schematic diagram of performing computations for a first set of classifiers in an embodiment of the invention;

FIG. 11 is a schematic diagram of the computation of the second set of classifiers and the third set of classifiers performed in an embodiment of the invention; and

FIG. 12 is a diagram illustrating the computation of the third set of classifiers in an embodiment of the invention.

Detailed Description

In addition, the embodiments of the present invention and the features of the embodiments may be combined with each other without conflict.

The present invention will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.

The embodiment of the invention provides a method for estimating a road surface friction coefficient, which can comprise the following steps of S10-S30:

in step S10, a plurality of sets of classifiers for calculating the road surface friction coefficient are determined.

Here, the classifier set is a functional module that classifies road surface friction capacities in real time and estimates the corresponding road surface friction capacities, respectively, wherein the road surface friction capacities are characterized by road surface values associated with friction coefficients, so that the embodiment of the present invention attempts to estimate the road surface friction capacities by calculating the road surface friction coefficients.

In an embodiment of the present invention, the set of classifiers is determined according to vehicle operation data and vehicle lateral control related data, and the vehicle operation data includes vehicle speed, lateral acceleration and Steering rate (Steering rate is derived from Steering angle) of the vehicle, and the vehicle lateral control related data refers to data associated with lateral control of the vehicle, which includes tire slip angle of the vehicle, tire lateral force, first Self-aligning Torque (SAT) estimated based on Electric Power Steering (EPS) system, second SAT estimated based on tire lateral dynamics model, and derivatives of mean and variance of the first SAT in sampling time, wherein each set of classifiers includes parameters and calculation models for calculating road surface friction coefficient different from other sets of classifiers, wherein the parameter is part of data comprised by the vehicle operation data and/or data comprised by the vehicle lateral control related data.

In a preferred embodiment, determining the plurality of sets of classifiers may comprise the steps of:

and step S11, acquiring the vehicle running data when the driver operates the vehicle in a side direction.

And step S12, detecting whether the vehicle runs in a straight line or not based on the vehicle running data, if so, setting a straight line mark to be 1, otherwise, setting the straight line mark to be 0. Only if the vehicle does not run straight, the lateral control such as steering, lane changing, and cornering is involved, and the calculation of the lateral road surface friction coefficient is involved.

Step S13, modeling the EPS system with an extended state observer to estimate the first SAT for the vehicle at each time instant when the straight line flag is set to 0, and calculating derivatives of the estimated mean and variance of the first SAT over the sample time.

Step S14, estimating a second SAT of the vehicle based on the vehicle operation data, the first SAT, and a tire lateral dynamics model of the vehicle. This step also involves the estimation of the tire pull torque, as will be described below.

Step S15, estimating a tire lateral force of the vehicle using the bicycle model, and estimating the tire slip angle based on a correspondence relationship between the tire lateral force and the tire slip angle.

Step S16, determining the first, second, third, and fourth sets of classifiers based on the acquired vehicle operation data, the calculated derivatives of the mean and variance of the first SAT, and the estimated first SAT and tire slip angle.

The specific classifier set determination scheme corresponding to steps S11-S16 will be described in detail below with reference to fig. 1, and will not be described herein again.

And step S20, selecting a corresponding classifier set according to the tire slip angle, the tire lateral force and the steering rate generated by the driver operating the vehicle laterally to calculate the road surface friction coefficient.

Wherein each set of classifiers is preconfigured to match a set of selection conditions including the tire slip angle, the tire lateral force, and the steering rate, such as selecting a preconfigured set of designated classifiers when the tire slip angle is greater than a certain threshold.

The specific classifier set selection scheme will be described in detail below with reference to fig. 1, and will not be described herein again.

In step S30, the calculated maximum road surface friction coefficient is determined as the final road surface friction coefficient.

For example, when a driver performs a lateral operation such as steering, lane changing, and cornering, different tire slip angles, tire lateral forces, and steering rates may occur due to different degrees (or forces) of the operation, so that a plurality of sets of classifiers may be selected to calculate a plurality of sets of road surface friction coefficients, and therefore, after the sets of classifiers satisfying conditions all calculate the road surface friction coefficients, the calculated maximum road surface friction coefficient needs to be determined as a final road surface friction coefficient, which may represent a lateral road surface friction capability, so as to achieve a purpose of estimating the maximum lateral friction capability before a vehicle actually reaches the maximum road surface friction capability.

FIG. 1 is a schematic diagram of a method module 100 for estimating a road surface friction coefficient in an embodiment of the invention, which estimates the road surface friction coefficient based on vehicle operation data acquired from an EPS system of a vehicle or the like. Therefore, a vehicle having an EPS system equipped with a steering column and a controller for determining a road surface friction coefficient in an embodiment of the invention will be described first with reference to fig. 2.

As shown in fig. 2, the vehicle 10 includes an EPS system 20 with a steering column mounted and a controller 50. The controller 50 is schematically shown as a single unit, however, various elements of the controller 50 may be distributed among a plurality of dedicated control units, such as a motor control unit, a steering control unit, etc., or Electronic Control Units (ECUs).

The controller 50 is configured to estimate a Lateral Friction Capability (LFC) of a road surface. The controller 50 is also configured to perform control actions appropriate for the determined LFC, such as by displaying information about the road surface condition to the driver of the vehicle 10 via the display 17 (e.g., display screen, indicator light, icon, etc.), or communicating the LFC of the road surface to other active chassis subsystems (not shown) to control one or more vehicle components to assist the driver in operating the vehicle in a safe manner.

The vehicle 10 includes a steering wheel 12. The steering wheel 12 rotates in response to a steering input by the driver. The steering wheel 12 is operatively connected to a steering column 14, and the steering column 14 is in turn connected to a steering mechanism 16. In one embodiment, the steering mechanism 16 is a rack and pinion assembly, although other assemblies may be used depending on the design. Such as those skilled in the art, the steering assembly 26 and steering mechanism 16 ultimately orient the tire 25 relative to the road surface 27 by moving the tire tie rod 18 on a set of front axles (not shown).

The torque sensor 23 and the steering angle sensor 21 may be positioned relative to the steering column 14. The torque sensor 23 measures and transmits a torque signal (arrow 123) to the controller 50. Likewise, the steering angle sensor 21 measures and transmits a steering angle signal (arrow 121) to the controller 50. The controller 50 processes the

signals

121, 123 and other vehicle operating data (arrow 11), such as vehicle speed, mass, etc., and determines the amount of auxiliary steering required by the steering motor 32 to perform the current steering operation. The controller 50 communicates with the steering motor 32 via motor control signals (arrow 13). The steering motor 32 generates and outputs a motor torque (arrow 15) through a reduction gear set 33 and the steering mechanism 16 in response to the motor control signal (arrow 13).

Still referring to fig. 2, the Controller 50 may transmit the motor control signal (arrow 13) to the steering motor 32 using a Controller Area Network (CAN), a serial bus, a data router, and/or other suitable Network connection.

In various embodiments, the sensor module 52, which may be disposed within the controller 50, directly transmits the signals required to perform the method module 100 for estimating the road surface coefficient of friction shown in FIG. 1. It should be noted that the sensor module 52 may be independent of the controller 50, and the data detected by the sensor module 52 may be transmitted to another controller (not shown) and then transmitted from the other controller to the method module 100 via a communication bus (not shown) or other communication device.

In various embodiments, as will be discussed in more detail below, the method module 100 determines the road surface type based on a plurality of road surface detection classification techniques that evaluate data obtained during a steering operation. The road surface may be, for example, an icy surface, a snowy surface, a dry surface, or other types. The method module 100 determines a road surface value, which characterizes the road surface friction capacity, based on the determined road surface type associated with the friction coefficient. The road surface value may be a nominal value, for example, between 0 and 1, which is associated with a particular road surface type. Typical pavement values for ice are 0.1-0.2, snow 0.3-0.4, wet 0.5-0.6, and dry 0.7-1.0. The method module 100 generates signals to control one or more components of the vehicle 10 based on the road surface values and/or road surface types and/or provides the road surface values and/or road surface types to other control systems (not shown) for further processing and control of the components of the vehicle 10.

The dynamics of the steering mechanism 16 can be modeled as:

wherein: theta_pIs a pinion angle, J_rpIs the inertia of the steering mechanism 16 (e.g., rack and pinion assembly), B_rpIs the damping coefficient, n is the gear ratio of the reduction gear set 33, C_fricIs the coulomb friction, T, acting on the steering rack of the steering assembly 26_aIs the assist torque (i.e., nT)_m)，T_mIs the motor torque, M_zIs SAT, J_mAnd B_mIs a steering motor32 corresponding inertia and damping coefficient, T_tsIs the output from the torque sensor 23. Torque applied by the driver of the vehicle 10, i.e. T_dWith assistance torque T from the EPS system 20_aThus, referring to equation (1), two reaction moments should be overcome when the vehicle 10 is turning: 1) SAT, i.e. M_zWhich is generated by the tires 25 and the road surface 27, and 2) the torque generated by the coulomb and viscous friction of the EPS system 20 itself.

Referring back to fig. 1, the method module 100 receives vehicle operation data 102 captured by the sensor module 52 (see fig. 2) including, but not limited to, pinion angle data, yaw rate data, longitudinal speed data, wheel speed, steering wheel angle data, motor torque data, torsion bar torque data, and lateral acceleration data. The self-aligning processing block 103 includes two parts, i.e., a process 104 of determining the occurrence of straight-ahead driving of the vehicle and a process 106 of estimating the first SAT, wherein fig. 3 is a flowchart of the process 104 of determining the occurrence of straight-ahead driving of the vehicle in the embodiment of the present invention, fig. 4 is a flowchart of the process 106 of estimating the first SAT in the embodiment of the present invention, and fig. 5 is a flowchart of calculating the derivative of the mean and variance of the first SAT in the embodiment of the present invention. Determining vehicle straight-ahead travel using vehicle operation data 102 with reference to FIG. 3 and using an extended state observer to obtain the first SAT (M) with reference to FIG. 4_zeps) May be calculated (as may be understood by reference to what is described in U.S. patent application No. 13/075,263 to the inventor of the present application, the contents of which may be incorporated herein), and to calculate the derivatives of the mean and variance of the first SAT with reference to fig. 5.

The evaluation module 107 comprises three parts, namely a process 108 of determining a linear region of lateral force dynamics, a process 110 of estimating a second SAT, and a process 112 of estimating a tire lateral force and a tire slip angle, wherein fig. 6 is a schematic flow diagram of the process 108 of determining a linear region of lateral force dynamics in an embodiment of the invention, fig. 7 is a schematic diagram of an exemplary tire lateral dynamics model in an embodiment of the invention, fig. 8 is a schematic diagram of tire lateral force (solid line) and SAT (dashed line) versus tire slip angle in an embodiment of the invention, and fig. 9 is a schematic diagram of a free body diagram of a bicycle model for a turning maneuver in an embodiment of the invention. The evaluation module 107 takes as input vehicle operation data such as steering angle, first SAT, vehicle speed, yaw rate, and lateral acceleration. Based on these inputs, the evaluation module 107 determines a linear region 108 of lateral force dynamics with reference to fig. 6, and with further reference to fig. 7 and 8, estimates the tire drag and the second SAT based on the tire lateral dynamics model as examples of information contained in the tire lateral dynamics model, and finally, with reference to fig. 9, estimates the tire lateral force and the tire slip angle.

The road friction classification module 117 includes a process 118 of performing a calculation of a set of classifiers, where fig. 10 is a schematic diagram of performing a calculation of a first set of classifiers, fig. 11 is a schematic diagram of performing a calculation of a second set of classifiers and a third set of classifiers, and fig. 12 is a schematic diagram of performing a calculation of a third set of classifiers, which will be described in detail below with respect to the first to fourth sets of classifiers. The road friction classification module 117 receives as input vehicle operating data including steering angle, vehicle speed, yaw rate, and lateral acceleration data. Additionally, road friction classification module 117 receives the linear assessment of process 108, the second SAT estimated by process 110, the first SAT estimated by process 106 and the derivatives of the calculated mean and variance, the tire slip angle estimated by process 112, based on which process 118 may, in a preferred embodiment, determine four different sets of classifiers that depend on tire slip angle, etc., with reference to fig. 10-12.

The four classifier sets are specifically as follows:

a first Set of classifiers (hereinafter Set1) including the first SAT and the tire slip angle (corresponding to the parameters for calculating the road surface friction coefficient referred to in step S10, and the other sets of classifiers being similar thereto), and calculating a first road surface friction coefficient based on the first SAT (corresponding to the calculation model for calculating the road surface friction coefficient referred to in step S10, and the other sets of classifiers being similar thereto);

a second Set of classifiers (hereinafter Set2) comprising the tire slip angle, a derivative of the variance of the first SAT over a sample time, a vehicle speed, and calculating a second road surface friction coefficient based on the derivative of the variance of the first SAT over the sample time;

a third Set of classifiers (hereinafter Set3) comprising the tire slip angle, a derivative of the average of the first SAT over a sample time, a vehicle speed, and a third road surface friction coefficient based on the derivative of the average of the first SAT over a sample time; and

a fourth Set of classifiers (hereinafter Set4) that includes the lateral acceleration and calculates a fourth road friction coefficient based on the lateral acceleration.

Therefore, the four classifier sets can be selected to calculate the corresponding road surface friction coefficient according to the current running state of the vehicle. In a preferred embodiment, for step S20, selecting a corresponding set of classifiers to calculate the road surface friction coefficient according to the tire slip angle, the tire lateral force and the steering rate generated by the driver operating the vehicle laterally may include: selecting Set1 to calculate the first road friction coefficient when the tire slip angle is greater than a Set slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is greater than a Set steering rate threshold; selecting Set2 to calculate the second road friction coefficient when the tire slip angle is less than the slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is greater than the steering rate threshold; selecting Set3 to calculate the third road surface friction coefficient when the tire slip angle is less than the slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is less than the steering rate threshold; when the tire lateral force indicates non-linearity, Set4 is selected to calculate the fourth road friction coefficient. Additionally, for a fourth road friction coefficient, a fourth set of classifiers may also be selected to calculate the fourth road friction coefficient when a difference between the first SAT and the second SAT is greater than a set threshold.

The decision block 121 includes a process 122 of determining a final road surface friction coefficient, which corresponds to the above-described step S30, and determines the maximum value among the calculated first to fourth road surface friction coefficients as the final road surface friction coefficient.

Details of the implementation of various portions of the method module 100 of fig. 1 are described in detail below in conjunction with fig. 3-12.

Detecting occurrence of straight-line running of vehicle

Fig. 3 schematically illustrates a process 300 for detecting the occurrence of straight-ahead driving of a vehicle in an embodiment of the invention. Whether the vehicle is running straight or not may be indicated by the state of the straight flag, and if the straight flag is 1, the vehicle is running straight (or moving), and if the straight flag is 0, the vehicle is not running straight. Table 1 shows the correspondence of the various function blocks numbered in fig. 3 with the corresponding functions, and with reference to fig. 3, the steps included in a process 300 for detecting the occurrence of straight-ahead running of a vehicle can be understood in conjunction with table 1.

TABLE 1

Where, at the start of detecting the occurrence of a straight-line running of the vehicle, the function block 304 calculates a plurality of differential wheel speeds (differential wheel speeds), including:

ΔV₁₁＝Abs(V_LF-V_RF)

ΔV₃₄＝Abs(V_LR-V_RR)

ΔV₁₄＝Abs(V_LF-V_RR)

ΔV₂₃＝Abs(V_RF-V_LR)

wherein, V_LFIs the left front wheel speed, V_RFIs the right front wheel speed, V_LRIs the right front wheel speed, V_RRIs the right rear wheel speed, which can be measured by the associated sensor.

Wherein the differential wheel speed represents a comparison of all left, right, front and rear wheel positions, and the differential wheel speed is compared to a differential threshold V_th1And V_th2Making a comparison, wherein a differential speed threshold V is_th1And V_th2Representing the maximum speed difference associated with vehicle operation on a straight line.

Second, relating to the first SAT

Referring to FIG. 4, a block diagram of one possible embodiment of an extended state observer corresponding to process 106 is shown. Referring again to FIG. 2, the extended state observer models the EPS system 20 to estimate the first SAT. Given a set of control inputs (u) and control outputs (y), state estimation is performed. Thus, the state (x) of the system may be at each time T and sample time T_s1(say 10 ms) modeled as, for example:

where (t) represents time, A, B, C and D are calibration values. The state observer model can then be derived as:

where L in this equation (3) is the estimator gain matrix. The above equations of state are readily understood by one of ordinary skill in the art.

Thus, the controller 50 estimates a first SAT value (denoted as M) using the EPS system (e.g., the steering mechanism described above) and using the extended state observer 106_zeps). The SAT estimator function is based on the model of the steering mechanism 16 shown in equation (1) above and may be mathematically represented as follows:

wherein the content of the first and second substances,

and u ═ i.

In these equations, w representsExternal disturbances, such as torque loads, for example, the difference between the torque value measured by the torque sensor 23 in fig. 2 (arrow 123 in fig. 2) and the road torque. Internal dynamics

In combination with the external disturbance w, a generalized disturbance can be formed

The above equation is then rewritten as:

wherein the content of the first and second substances,

then, the controller 50 may derive a state space model of the enhanced canonical (augmented canonical) as follows with reference to fig. 3:

C＝[1 0 0],D＝[0]

wherein the content of the first and second substances,

including the interference to be estimated.

Next, the state space model from the extended state observer 106 is discretized by applying zero-order hold. At each instant T and sampling time T:

using the extended state observer 106, the following results are obtained:

by definition L ═ Φ L_cThe estimated value is reduced to:

therein, a new state, i.e. a discrete estimator, is given by:

the SAT estimator gain vector L is then determined by placing the pole (ζ) of the discrete feature equation λ (z) as follows_c：

λ(z)＝|zI-(Φ-ΦLH)|＝(z-ζ)³(sayζ＝0.002) (11)

Apply zero order hold again:

H＝[1 0 0]j is 0, and

thereby:

third, calculating derivatives of the mean and variance of the first SAT

FIG. 5 schematically continues the flow of FIG. 4 (and also corresponds to process 106) for computing a mean and a square of the first SATThe derivative of the difference. Calculating an intermediate value M of the mean and variance of the first SAT at each sampling time i_iAnd S_iThe average value is calculated as (M)_iI) and then calculating the variance as (S)_iI) and then calculating the derivative Mean _ M of the Mean of the first SAT_{zeps_dot}Derivative of sum variance Var _ M_{zeps_dot}. Table 2 shows the correspondence of the numerically labeled functional blocks of fig. 5 to the corresponding functions, and with reference to fig. 5, the steps of calculating the derivative of the mean and variance of the first SAT can be understood in conjunction with table 2.

TABLE 2

Fourthly, determining the linear area of the lateral dynamics of the tire

Fig. 6 schematically illustrates a process 108 for determining a linear region of tire lateral dynamics. Table 3 shows the correspondence of the various functional blocks indicated by numbers in fig. 6 with the corresponding functions, and with reference to fig. 6, the step of determining the linear region of the lateral dynamics of the tyre can be understood in connection with table 3.

TABLE 3

Fifth, tire lateral dynamics model and second SAT

Referring to fig. 7, an example of the information contained in the tire lateral dynamics model is schematically shown (corresponding to process 108). The vertical axis 61 represents the size, and the horizontal axis 63 represents the traveling direction of the vehicle 10 shown in fig. 2. The lateral force acting on a given tire 25 is represented by arrow 64 and the tire contact width (arrow 76) is represented by the area between dashed line 71 and dashed line 73 shown in dashed-line format. The additional quantities shown in FIG. 7 include bitsA tire slip zone (arrow 60) between

points

70 and 72, a tire attachment zone (arrow 62) between

points

72 and 74, a tire trailing (arrow 65), a vehicle heading (arrow 68), and a tire contact length (arrow 78), wherein the tire trailing (arrow 65) is the area beginning with the lateral force (arrow 64) and ending at point 77. Arrow 75 indicates SAT (i.e., M)_z). The slip angle (α) of the front tire 25 is between the direction of travel (arrow 63) and the heading (arrow 68) (i.e., the orientation of the tire 25). The linear region of the tire lateral dynamics model is generally indicated by arrow 80. The SAT characteristics are explained based on the lateral force distribution of the tire contact patch, which is the portion of the tire 25 in contact with the road surface 27, represented by the double arrow 78. Lateral forces (arrows 64) accumulate in this contact surface to a point 72 at which tread shear overcomes the available frictional forces. This is the attachment area indicated by double arrow 62. When the tire 25 of fig. 2 is rotated at slip angle (α), slip then occurs in the slip region (double arrow 60).

The asymmetric force distribution results in the point of application of the lateral force (arrow 64) being positioned toward the rear of the contact patch by a tire trailing distance (double arrow 65). As is known in the art, the term tire drag refers to the distance from the center of the tire 25 to the point where the lateral force is formed. In other words, the asymmetric lateral force distribution caused by the stick/slip condition affects the tire footprint (double arrow 65). Thus, SAT variation indicates a stick/slip condition in the tire contact patch, since SAT is equal to the lateral force (arrow 64) multiplied by the tire traction distance (double arrow 65). However, as shown in fig. 8, SAT reaches its maximum well before the lateral forces saturate.

Still referring to fig. 8, tire lateral dynamics is indicated by arrow 220 and SAT is indicated by arrow 230. Tire lateral dynamics curves and SAT curves lateral tire slip was measured along the vertical axis (arrow 200) and horizontal axis (arrow 260) shown in fig. 8. Significant difference between the tire lateral dynamics curve 220 and the SAT 230. The peak (arrow 210) of the SAT at lateral tire slip (arrow 250) is less than the peak (arrow 270) of the tire lateral dynamics at lateral tire slip (arrow 280). Similarly, there are other linear relationships, namely, the initial slope indicated by arrow 290 of the SAT curve 230 ends at tire slip (arrow 300), which is much less than the initial slope indicated by arrow 240 of the tire lateral dynamics (which ends at lateral tire slip (arrow 310) and lateral force (arrow 380)).

Due to M_Z＝-L_pF_yfAnd since the SAT reaches its peak and begins to decay before the lateral force, the tire track cannot be assumed constant and needs to be estimated.

Returning to FIG. 7, the controller 50 of FIG. 2 thus applies a force F in the tire lateral direction_yUsing a tire lateral dynamics model (corresponding to process 110) to estimate tire drag in a linear region (arrow 80) of (a) where slip angle α and low frequency self-aligning component M_zAnd (4) in proportion.

So that:

M_z＝-L_pF_yf,F_yf＝C_fα_f (13)

wherein L is_pIs the tire drag (arrow 65 of FIG. 7), C_fIs the cornering stiffness of the front tire 25, F_yfIs the front tire lateral force, and α_fIs the front tire slip angle, wherein α_fThe following can be calculated:

wherein delta_rIs the front wheel steering angle (i.e. steering wheel angle divided by steering gear ratio), v_yIs the lateral velocity, v, of the vehicle at the center of gravity_xIs the longitudinal speed of the center of gravity,

is the yaw rate of the vehicle 10, and a is the distance from the center of gravity of the vehicle 10 to its front axle.

In the linear region (arrow 80), the lateral velocity (v)_y) From the bicycle model equation (15) and the kinematics equation (16), the following can be calculated:

where b is the distance from the center of gravity of the vehicle 10 to the rear axle, C_rThe cornering stiffness (not shown) of the two tires which are the rear axles, g is the gravitational acceleration, m is the vehicle total weight, and γ is the road bank angle, i.e. the inclination of the road surface 27 of fig. 2 can be estimated from information such as the lateral acceleration and yaw rate.

From the above equation, the controller 50 may calculate the lateral velocity v as follows_yAnd thus calculates the second SAT (i.e., M)_zdyn)：

Wherein the content of the first and second substances,

and

it should be noted that the tire cornering stiffness C_fAnd C_rAre known values and may be provided by the tire manufacturer. On the other hand, the tire trailing distance L_pIs unknown and can be estimated by reference to the above.

SAT M determined in equation (12)_zepsUsed in equation (18) to estimate tire drag L in the linear region of tire lateral dynamics_p. On the other hand, when the tire is fully slipped, both the tire footprint and the SAT tend to zero, which prevents the use of self in the lateral friction estimationAnd (4) restoring the torque. In this case, different methods will be used for determining the maximum lateral friction coefficient of the road surface, which will be described in detail below and will not be described further here.

Equation (18) can be expressed as follows

Wherein the content of the first and second substances,

in equation (19), at each time T and sampling time T_s1(e.g., 10 milliseconds) y and φ are known, and θ is unknown. An estimation model including a least squares estimation is performed and described in equation (20) as follows.

Wherein:

the error term epsilon is associated with the error in the estimated tire footprint. The term P is in embodiments of the present invention interpreted as the covariance of the selected parameter, which has a magnitude that provides a measure of the uncertainty of the parameter value.

The initial value of the tire slip θ will be set to L within the linear range of self-aligning torque (arrow 300) defined by tire slip_p0. Assuming that the tire trailing is constant, L is L for an exemplary tire of tire width 235, aspect ratio 55, and rim diameter 19 (described as R235/55R19)_p0An exemplary value of (d) is 0.06 m.

This variable k is called a weighting factor, the calculation of which is described below.

Example value (κ)₀＝0.05)(21)

If a sudden change in tire traction occurs when the SAT crosses its peak, the estimation error ε²The square of (t) increases and p (t) increases rapidly so that rapid adaptation can occur. After adaptation, the error ε²(t) decreases and κ should return to near 1.

Converting the estimation model shown in equation (20) to be periodically executed to determine

An algorithm of (1), wherein

Performing the estimation model using least squares estimation as described herein results in estimating the tire footprint and the second SAT, i.e.

It should be noted that the first SAT M estimated in process 106_zepsAnd the SAT M estimated above_zdynAre equal until the SAT and tire footprint approach zero, in which case the estimation algorithm in equation (20) will result in large errors. Tire trailing distance L_pAnd the first SAT M_zepsAnd a second SAT M_zdynThe error between will increase indicating that the tire is entering a nonlinear region of its lateral dynamics. This information will be used to determine the final friction coefficient of the road surface.

Sixth, tire side force and tire offset angle

Referring to FIG. 9, a free body diagram of a bicycle model in a turning maneuver is shown. The origin of the vehicle axle system is fixed at the center of gravity of the vehicle. The vehicle axis is oriented right-hand, with the x-axis pointing forward, the y-axis pointing right, and the z-axis pointing downward (arrow 780). During a turning maneuver, the vehicle has a longitudinal direction (arrow 790), a lateral direction (arrow 800) and an angular velocity (arrow 730), as shown in FIG. 9. Since the bicycle model assumes a constant forward speed of the vehicle, there is no longitudinal acceleration, so the model can be reduced to two degrees of freedom. The remaining motion variables are lateral velocity (arrow 800) and angular velocity (yaw rate) about the z-axis (arrow 730). The equations of motion can be derived from the free body diagram of fig. 9. The left side of the equation of motion contains the resultant lateral and longitudinal forces applied to the vehicle by the tires, as well as the yaw moment. The longitudinal force applied to the vehicle by the tire is zero due to the absence of longitudinal acceleration. Therefore, the equation of motion can be derived in the following form.

Wherein, F_yf(arrow 720) and F_yr(arrow 750) is the front and rear lateral tire force and is the only unknown variable. The vehicle parameters (m, a, b and yaw inertia I) can be performed by simple measurements_z). Lateral acceleration (a)_y) And yaw rate

(arrow 730) is the measurable signal obtained from the sensor.

Thus, the axis-specific lateral tire force comes from equation of motion (21).

There is a point in the turning maneuver called the instant Center of Gravity (CG) (arrow 820) around which the vehicle moves. The position of this point is defined by the lateral slip angle (β) (arrow 810) and the front tire slip angle (α) of the vehicle, respectively_f) (arrow 710) and rear tire slip angle (α)_r) (arrow 740) is determined. These three angles are defined by the ratio of the lateral and longitudinal velocities of the location in question. It must be noted, however, that the front tires are tilted by the steering angle (δ r). Thus, the tire slip angle (α) relative to the heading direction of the tire_fArrow 710) is the steering angle (δ r, arrow 700) minus the lateral velocity (arrow)800) And longitudinal velocity (arrow 810). Referring again to fig. 8, a typical relationship between tire lateral slip angle (arrow 260) and lateral force (arrow 220) is given by the following equation for the front and rear tires.

Wherein C is_fAnd C_rIs the stiffness of the front and rear tires. In the embodiment of the present invention, only the front axle (steerable axle) is used to determine the lateral force of the road surface. From equation (22), the front-to-side force F can be determined_yfThus, the front wheel slip angle (α) can be estimated_f). It can be seen in fig. 8 that at the lateral slip angle (arrow 310), the tire stiffness is constant to a good approximation. This condition is referred to as the linear operating range of the vehicle. At the lateral slip angle above arrow 310 and before arrow 280, the linear approximation is no longer valid, i.e., the vehicle reaches a non-linear operating range. In order to accurately estimate the tire slip angle in a non-linear range, it is necessary to adjust the tire stiffness to account for the non-linear operating range.

At the lateral slip angle above arrow 280, the lateral force saturates and the tire stiffness is effectively zero. In this case, it can be assumed that the vehicle slips.

Tire stiffness will be in terms of linear tire drag L_p0Estimated tire drag

And vehicle speed v_xThe following adjustments were made:

C_f＝C_f0f₁(L_p,L_p0)f₂(v_x) (24)

(R235/55R19) tire (combination of two front tires) C_f0174000N/rad (which may be measured or provided by the tire supplier).

Exemplary values for f1 and f2 are as follows:

L_p0/L_p	0.003	0.019	0.058	0.123	0.214	0.33	0.47	0.63	0.87	1
											f₁	0.1	0.2	0.3	0.4	0.5	0.6	0.7	0.8	0.9	1

it is obvious that when L is_p＝L_p0The correction factor is equal to 1, so that the following table can be obtained.

v _x(kph)	10	30	40	50	80	100	150
								f₂	1.05	1.05	0.83	0.83	0.7	0.6	0.52

Tire forward slip angle (arrow 710)) may then be defined as follows:

seven, calculation of classifier set

Referring back to fig. 1, road friction classification module 117 receives the linear evaluation indicator (process 108), estimates tire drag distance (process 110), the first and second SATs, derivatives of the mean and variance of the first SAT, the estimated tire slip angle (process 112), and the vehicle straight ahead indicator (process 104). Based on these data, process 118, which performs the calculation of the Set of classifiers, may determine four different sets of classifiers, i.e., Set1-Set4 described above, based on the values of tire slip angles. Fig. 10 schematically illustrates the calculation of Set1 in process 118 of road friction classification module 117. Table 4 shows the correspondence of the respective function blocks marked with numerals in fig. 10 with the corresponding functions, and referring to fig. 10, the step of performing the calculation of Set1 can be understood in conjunction with table 4.

TABLE 4

Wherein, f1(max (M)_zeps) Is the relationship that maps the maximum SAT achieved on a given road surface to the maximum lateral coefficient of friction of the road surface. The map is obtained by collecting data from different road surfaces having different known lateral coefficients of friction. Exemplary values for the first Set of classifiers (Set1) for an exemplary tire (R235/55R19) are shown in the following table:

maximum SAT (Nm)	23	48	85	115	150	215	300
								Maximum lateral coefficient of friction (g)	0.1	0.2	0.3	0.4	0.5	0.7	1

Fig. 11 schematically illustrates the calculation of Set2 and Set3 in process 118 of road friction classification module 117. Table 5 shows the correspondence of the respective function blocks marked with numerals in fig. 11 with the corresponding functions, and referring to fig. 11, the steps of performing the calculations of Set2 and Set3 can be understood in conjunction with table 5.

TABLE 5

f2(max(Var_M_{zeps_dot}，v_x) And f3(max (Mean _ M)_{zeps_dot}Vx)) is a relationship that maps the maximum of the derivative of the mean and variance of the first SAT achieved on a given road surface to the maximum lateral friction coefficient of the road surface. Corresponding relational tables can be obtained by collecting data from different road surfaces having different known lateral coefficients of friction. Exemplary values for Set2 and Set3, defined as exemplary tires (R235/55R19), are shown in the table below.

For Set2, where horizontal rows represent vehicle speed v_x(in kph), max (Var _ M) in vertical alignment_{zeps_dot}

	20	30	40	50	60
						0	0.1	0.1	0.1	0.1	0.1
5	0.15	0.1	0.1	0.1	0.1
						10	0.2	0.15	0.1	0.1	0.1
15	0.2	0.2	0.1	0.1	0.1
						20	0.2	0.3	0.15	0.1	0.1
25	0.25	0.4	0.15	0.1	0.1
						30	0.25	0.7	0.2	0.1	0.15
35	0.3	0.7	0.25	0.2	0.15
						40	0.35	0.7	0.3	0.2	0.2
45	0.4	0.7	0.4	0.25	0.1
						50	0.7	0.7	0.7	0.3	0.15
55	0.7	0.7	0.6	0.35	0.15
						60	0.7	0.7	0.7	0.4	0.3
65	0.7	0.7	0.9	0.45	0.3
						70	0.7	0.7	0.9	0.5	0.35
75	0.7	0.7	0.9	0.7	0.35
						80	0.7	0.7	0.9	0.8	0.4
85	0.7	0.7	0.9	0.9	0.45
						90	0.7	0.7	0.9	1	0.5

For Set3, where horizontal rows represent vehicle speed v_x(in kph), and max (Mean _ M) in vertical alignment_{zeps_dot})

	20	30	40	50	60	80	100
								5	0.1	0.1	0.1	0.1	0.1	0.1	0.1
10	0.1	0.1	0.1	0.1	0.1	0.1	0.1
								15	0.15	0.15	0.15	0.15	0.15	0.15	0.15
20	0.15	0.15	0.15	0.15	0.15	0.15	0.15
								25	0.2	0.2	0.2	0.2	0.2	0.2	0.2
30	0.3	0.25	0.25	0.25	0.25	0.25	0.25
								35	0.4	0.25	0.25	0.25	0.25	0.25	0.25
40	1	0.3	0.3	0.3	0.3	0.3	0.3
								45	1	0.4	0.3	0.3	0.3	0.3	0.3
50	1	0.5	0.35	0.35	0.35	0.35	0.35
								55	1	0.6	0.35	0.35	0.35	0.35	0.35
60	1	1	0.4	0.4	0.4	0.4	0.4
								65	1	1	0.5	0.45	0.45	0.45	0.45
70	1	1	0.7	0.45	0.45	0.45	0.45
								75	1	1	1	0.5	0.5	0.5	0.5
80	1	1	1	0.6	0.55	0.55	0.55
								85	1	1	1	0.7	0.6	0.6	0.6
90	1	1	1	0.9	0.7	0.65	0.65
								95	1	1	1	1	0.8	0.65	0.65
100	1	1	1	1	0.9	0.7	0.7
								105	1	1	1	1	1	0.9	0.9

Among them, when the driver's operation tendency is mild so that the steering excitation level is low, the method of estimating the road surface maximum friction coefficient by Set2 and Set3 is novel and highly applicable. Lower levels of steering excitation produce very low tire slip angles, making it very difficult to estimate road surface friction coefficients using SAT because SAT at low slip levels is not different from road surface to road surface. Thus, Set2 and Set3 overcome this drawback by using derivatives of the mean and variance of the SAT, so that the road surface maximum friction coefficient can be accurately estimated even if the steering excitation level is low.

Fig. 12 schematically illustrates the calculation of Set4 in process 118 of road friction classification module 117. Table 6 shows the correspondence of the respective function blocks labeled with numerals in fig. 12 with the corresponding functions, and referring to fig. 12, the step of performing the calculation of Set4 can be understood in conjunction with table 6.

TABLE 6

Eight, final road surface friction coefficient

Finally, decision module 121 receives U (1), U (2), U (3), and U (4) from road friction classification module 117, and process 122 will eventually maximize the lateral road friction coefficient μ_{max_lat}The determination is as follows:

μ_{max_lat}＝max(U(1)，U(2)，U(3)，U(4))

in summary, the conventional road surface friction coefficient estimation function can accurately estimate the maximum road surface friction coefficient on the premise that the driver adopts violent lateral operation, and if the operation trend of the driver is mild, the estimated coefficient value is distorted.

Based on the same inventive concept as the embodiment of the method for estimating a road surface friction coefficient, the embodiment of the present invention further provides an estimation device of a road surface friction coefficient, including: the classifier determining module is used for determining a plurality of classifier sets used for calculating the road surface friction coefficient; the classifier selection module is used for selecting a corresponding classifier set to calculate the road surface friction coefficient according to the tire slip angle, the tire lateral force and the steering rate generated by the driver performing lateral operation on the vehicle; and the lateral friction coefficient determining module is used for determining the calculated maximum road friction coefficient as a final road friction coefficient.

Wherein the set of classifiers is determined from vehicle operational data comprising a vehicle speed, a lateral acceleration and a steering rate of the vehicle and vehicle lateral control related data comprising a tire slip angle of the vehicle, a tire lateral force, a first self-aligning torque (SAT) estimated based on an electric power steering system, a second SAT estimated based on a tire lateral dynamics model and derivatives of mean and variance of the first SAT over a sample time, wherein each set of classifiers comprises parameters and computational models for calculating road surface friction coefficients different from the other sets of classifiers, wherein the parameters are part of data comprised by the vehicle operational data and/or data comprised by the vehicle lateral control related data.

Wherein each set of classifiers is preconfigured to match a set of selected conditions comprising the tire slip angle, the tire lateral force, and the steering rate.

In a preferred embodiment, the classifier determination module comprises: the vehicle operation data acquisition submodule is used for acquiring the vehicle operation data when a driver performs lateral operation on the vehicle; a first SAT estimation sub-module, configured to detect whether a vehicle is traveling straight based on the vehicle operation data, set a straight flag to 1 if the vehicle is traveling straight, set the straight flag to 0 if the vehicle is traveling straight, model the EPS system using an extended state observer to estimate a first SAT of the vehicle at each time when the straight flag is set to 0, and calculate a derivative of a mean and a variance of the estimated first SAT during the sampling time; a second SAT estimation sub-module for estimating a second SAT of the vehicle based on the vehicle operating data, the first SAT, and a tire lateral dynamics model of the vehicle; a tire slip angle estimation sub-module for estimating a tire lateral force of a vehicle using a bicycle model and estimating the tire slip angle based on a correspondence relationship of the tire lateral force and the tire slip angle; and a set determination sub-module for determining the first, second, third and fourth sets of classifiers based on the acquired vehicle operation data, the calculated derivatives of the mean and variance of the first SAT, and the estimated first SAT and tire slip angle.

The implementation details and the effects of the device for estimating the road surface friction coefficient of the embodiment of the invention are the same as or similar to those of the method for estimating the road surface friction coefficient, and are not repeated herein.

Embodiments of the present invention also provide a machine-readable storage medium having instructions stored thereon, where the instructions are used to enable a machine to perform the above-mentioned method for estimating a road surface friction coefficient. The machine-readable storage medium includes, but is not limited to, phase change Memory (PRAM), Static Random Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read Only Memory (ROM), Electrically Erasable Programmable Read Only Memory (EEPROM), Flash Memory (Flash Memory) or other Memory technology, compact disc read only Memory (CD-ROM), Digital Versatile Disks (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, and the like, which can store program code.

An embodiment of the present invention further provides a vehicle, as shown in fig. 2, where the vehicle 10 includes: a steering wheel configured to rotate in response to a driver steering input, wherein the steering input includes a steering torque and a steering angle; a torque sensor configured to measure a steering torque; a steering angle sensor configured to measure a steering angle; the steering mechanism realizes follow-up with the steering wheel through a steering column; the machine-readable storage medium described above; a controller configured to receive vehicle operation data including the steering torque and the steering angle and execute instructions stored in the machine-readable storage medium based on the vehicle operation data.

The specific structure of the vehicle 10 can be understood by referring to the description about fig. 2, and the embodiment of the present invention is not described herein again.

The above description is only for the purpose of illustrating the preferred embodiments of the present invention and is not to be construed as limiting the invention, and any modifications, equivalents, improvements and the like that fall within the spirit and principle of the present invention are intended to be included therein.

Claims

1. A method of estimating a road surface friction coefficient, the method comprising:

determining a plurality of sets of classifiers for calculating a road surface friction coefficient, wherein the sets of classifiers are determined based on vehicle operating data and vehicle lateral control related data, and the vehicle operation data comprises vehicle speed, lateral acceleration and steering rate of the vehicle, the vehicle lateral control related data comprises tire slip angle, tire lateral force of the vehicle, first self-aligning torque estimated based on the electric power steering system, second self-aligning torque estimated based on the tire lateral dynamics model, and derivatives of mean and variance of the first self-aligning torque over a sampling time, wherein each classifier set comprises parameters and calculation models for calculating the road surface friction coefficient different from other classifier sets, wherein the parameter is part of data comprised by the vehicle operation data and/or data comprised by the vehicle lateral control related data;

selecting a corresponding set of classifiers to calculate road surface friction coefficients as a function of the tire slip angle, the tire lateral force, and the steering rate resulting from a driver's lateral operation of a vehicle, wherein each set of classifiers is preconfigured to match a set of selection conditions comprising the tire slip angle, the tire lateral force, and the steering rate; and

and determining the calculated maximum road surface friction coefficient as a final road surface friction coefficient.

2. The method of estimating a road surface friction coefficient according to claim 1, wherein the plurality of sets of classifiers for calculating a road surface friction coefficient include:

a first set of classifiers comprising the first self-aligning torque and the tire slip angle and calculating a first road friction coefficient based on the first self-aligning torque;

a second set of classifiers comprising the tire slip angle, a derivative of the variance of the first self-aligning torque over a sample time, a vehicle speed, and a second road surface friction coefficient based on the derivative of the variance of the first self-aligning torque over a sample time;

a third set of classifiers comprising the tire slip angle, a derivative of the average of the first self-aligning torque over a sample time, vehicle speed, and a third road friction coefficient based on the derivative of the average of the first self-aligning torque over a sample time; and

a fourth set of classifiers comprising the lateral acceleration and calculating a fourth road friction coefficient based on the lateral acceleration.

3. The method of claim 2, wherein the selecting a corresponding set of classifiers to calculate the road friction coefficient according to the tire slip angle, the tire lateral force and the steering rate caused by the driver's lateral operation of the vehicle comprises:

selecting a first set of classifiers to calculate the first road friction coefficient when the tire slip angle is greater than a set slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is greater than a set steering rate threshold;

selecting a second set of classifiers to calculate the second road friction coefficient when the tire slip angle is less than the slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is greater than the steering rate threshold;

selecting a third set of classifiers to calculate the third road friction coefficient when the tire slip angle is less than the slip angle threshold, the tire lateral force is indicative of linearity, and the steering rate is less than the steering rate threshold;

selecting a fourth set of classifiers to calculate the fourth road friction coefficient when the tire lateral force indicates a non-linearity.

4. The method of estimating a road surface friction coefficient according to claim 3, wherein the selecting a corresponding set of classifiers to calculate a road surface friction coefficient further comprises:

selecting a fourth set of classifiers to calculate the fourth road friction coefficient when a difference between the first self-aligning torque and the second self-aligning torque is greater than a set threshold.

5. The method of estimating a road surface friction coefficient according to claim 2, wherein the determining a plurality of sets of classifiers for calculating a road surface friction coefficient includes:

acquiring vehicle operation data when a driver performs lateral operation on a vehicle;

detecting whether the vehicle runs in a straight line or not based on the vehicle running data, if so, setting a straight line mark to be 1, otherwise, setting the straight line mark to be 0;

modeling the electric power steering system with an extended state observer to estimate the first self-aligning torque of the vehicle at each instant when the straight-line flag is set to 0, and calculating derivatives of the mean and variance of the estimated first self-aligning torque over the sampling time;

estimating a second self-aligning torque of the vehicle based on the vehicle operating data, the first self-aligning torque, and a tire lateral dynamics model of the vehicle;

estimating a tire lateral force of a vehicle using a bicycle model and estimating the tire slip angle based on a correspondence of the tire lateral force to the tire slip angle;

determining the first, second, third, and fourth set of classifiers based on the acquired vehicle operation data, the calculated derivatives of the mean and variance of the first aligning torque, and the estimated first aligning torque and tire slip angle.

6. An estimation device of a road surface friction coefficient, characterized by comprising:

a classifier determination module for determining a plurality of sets of classifiers for calculating a road surface friction coefficient, wherein the set of classifiers is determined based on vehicle operation data and vehicle lateral control related data, and the vehicle operation data comprises vehicle speed, lateral acceleration and steering rate of the vehicle, the vehicle lateral control related data comprises tire slip angle, tire lateral force of the vehicle, first self-aligning torque estimated based on the electric power steering system, second self-aligning torque estimated based on the tire lateral dynamics model, and derivatives of mean and variance of the first self-aligning torque over a sampling time, wherein each classifier set comprises parameters and calculation models for calculating the road surface friction coefficient different from other classifier sets, wherein the parameter is part of data comprised by the vehicle operation data and/or data comprised by the vehicle lateral control related data;

a classifier selection module for selecting a corresponding set of classifiers to calculate road surface friction coefficients according to the tire slip angle, the tire lateral force and the steering rate resulting from a driver's lateral operation of a vehicle, wherein each set of classifiers is preconfigured to match a set of selection conditions comprising the tire slip angle, the tire lateral force and the steering rate; and

and the lateral friction coefficient determining module is used for determining the calculated maximum road friction coefficient as the final road friction coefficient.

7. The apparatus for estimating a road surface friction coefficient according to claim 6, wherein the plurality of sets of classifiers for calculating a road surface friction coefficient include:

8. The apparatus for estimating a road surface friction coefficient according to claim 7, wherein the classifier selection module is further configured to:

9. The apparatus for estimating a road surface friction coefficient according to claim 8, wherein the classifier selection module is further configured to:

10. The apparatus for estimating a road surface friction coefficient according to claim 7, wherein the classifier determining module includes:

the vehicle operation data acquisition submodule is used for acquiring the vehicle operation data when a driver performs lateral operation on the vehicle;

a first self-aligning torque estimation submodule, configured to detect whether a vehicle is traveling in a straight line based on the vehicle operation data, set a straight line flag to 1 if the vehicle is traveling in a straight line, or set the straight line flag to 0 if the vehicle is traveling in a straight line, and model the electric power steering system with an extended state observer to estimate a first self-aligning torque of the vehicle at each time when the straight line flag is set to 0, and calculate a derivative of a mean value and a variance of the estimated first self-aligning torque within the sampling time;

a second self-aligning torque estimation sub-module for estimating a second self-aligning torque of the vehicle based on the vehicle operating data, the first self-aligning torque, and a tire lateral dynamics model of the vehicle;

a tire slip angle estimation sub-module for estimating a tire lateral force of a vehicle using a bicycle model and estimating the tire slip angle based on a correspondence relationship of the tire lateral force and the tire slip angle; and

a set determination submodule for determining the first, second, third and fourth sets of classifiers based on the acquired vehicle operation data, the calculated derivatives of the mean and variance of the first self-aligning torque, and the estimated first self-aligning torque and tire slip angle.

11. A machine-readable storage medium having stored thereon instructions for causing a machine to perform the method of estimating a road surface friction coefficient according to any one of claims 1 to 5.

12. A vehicle, characterized in that the vehicle comprises:

a steering wheel configured to rotate in response to a driver steering input, wherein the steering input includes a steering torque and a steering angle;

a torque sensor configured to measure a steering torque;

a steering angle sensor configured to measure a steering angle;

the steering mechanism realizes follow-up with the steering wheel through a steering column;

the machine-readable storage medium of claim 11;

a controller configured to receive vehicle operation data including the steering torque and the steering angle and execute instructions stored in the machine-readable storage medium based on the vehicle operation data.