CN110435623B - Automatic hierarchical automatic emergency braking control system of electric motor car of adjustment - Google Patents

Abstract

The invention relates to an automatic-adjusting classification automatic emergency braking control system for an electric vehicle. The method comprises the following steps: the system comprises vehicle-mounted distance measurement and speed measurement sensing equipment, a grading early warning control system, a safe distance calculation model, a vehicle inverse longitudinal dynamics calculation model, a hydraulic braking force and regenerative braking force distribution calculation module, a hydraulic braking system inverse model, an ESC and Booster active pressurization hydraulic pressure distribution module and a road surface information estimation model, improves the comfort when an AEB system is triggered, reduces the potential safety hazards of rear-end collision and the like caused by sudden and large deceleration of a vehicle, estimates the gradient, the adhesion coefficient and other information of the current driving road and the front road of the vehicle, controls the online adjustment of parameters, enhances the adaptation degree of the AEB system to different road surface conditions, fully recovers the braking energy, improves the endurance mileage, and fully exerts the advantages of an ESC and an electronic mechanical brake Booster of a vehicle body stability control system on active pressurization.

Description

Automatic hierarchical automatic emergency braking control system of electric motor car of adjustment

Technical Field

The invention relates to an automatic emergency braking control system of an electric vehicle, in particular to an automatic-adjusting classification automatic emergency braking control system of the electric vehicle.

Background

In recent years, with the continuous development of global automobile technology, the automobile holding amount is increased, and the vehicle traffic accidents are increased. In order to reduce the number of road traffic accidents, various automobile active safety technologies and passive safety technologies have been rapidly developed. As an important active safety technology for automobiles, an Automatic emergency brake System (AEB) can effectively avoid a large number of traffic accidents, and has gradually become a vehicle standard.

The AEB system mainly comprises an information acquisition and control system and an execution mechanism. The information acquisition mainly comprises the steps of monitoring the surrounding environment in real time through sensors such as radars and the like, and transmitting target object information to a control system. After receiving the target object information, the control system decides information such as expected deceleration through a control strategy according to the motion state of the vehicle and sends the information to an execution mechanism. The executing mechanism carries out corresponding operation on the command of the control system through the electronic throttle valve and the brake, thereby achieving the effects of avoiding collision or lightening collision.

However, there are some problems with current AEB systems, such as:

1. the existing vehicle-mounted AEB system control algorithm is in consideration of safety, and generally chooses to directly brake the vehicle at the maximum braking deceleration when the AEB is required to intervene, so that the comfort of the members is seriously influenced, and the danger of rear-end collision and the like of the vehicle in emergency braking at high speed is easy to occur.

2. The AEB system control algorithm does not comprise a function of identifying road surface information, and only structural parameters of a vehicle and kinematic parameters of the vehicle are considered in the design of the algorithm, so that the influence of different roads on the implementation effect of the AEB algorithm is considered.

3. The AEB system control algorithm does not have adaptivity to changes in road conditions, for example, when a vehicle runs on a road with a low adhesion coefficient, the maximum braking deceleration of the vehicle may not reach the preset value of the AEB system due to the decrease of the adhesion coefficient, thereby causing collision; when a vehicle runs on a road with a certain gradient, the actual braking force of the vehicle can be changed due to the resistance of the gradient, and if the AEB algorithm is not adjusted according to the gradient of the road surface, the vehicle which goes up the slope can be decelerated too early to bring distrust to a driver or the vehicle which goes down the slope can be decelerated insufficiently to cause collision.

4. For an electric vehicle carrying a partially decoupled or non-decoupled electric power-assisted brake system, when the AEB system is triggered and starts to apply a braking force to the vehicle, the brake pedal will automatically move under the drive of the power-assisted motor due to the mechanical connection between the brake pedal of the driver and the brake master cylinder, on one hand, discomfort and confusion of the driver can be caused, and on the other hand, when the driver wants to autonomously apply a larger braking force to the vehicle due to the fact that the brake pedal is not at the initial position, the time for the foot of the driver to move to the brake pedal is longer than usual, and the time difference can cause the driver to hesitate or operate mistakenly, so that the braking force is applied insufficiently and danger occurs.

Disclosure of Invention

In order to solve the technical problems, the invention provides an electric vehicle grading automatic emergency braking control system based on road information automatic identification and control parameter online adjustment for an electric vehicle carrying a non-decoupling or partially decoupling electric power-assisted braking system.

The invention relates to an automatic-adjustment electric vehicle classification automatic emergency braking control system, which comprises: the system comprises vehicle-mounted distance and speed measuring sensing equipment, a graded early warning control system, a safe distance calculation model, a vehicle inverse longitudinal dynamics calculation model, a hydraulic braking force and regenerative braking force distribution calculation module, a hydraulic braking system inverse model, an ESC and Booster active pressurization hydraulic pressure distribution module and a road surface information estimation model, wherein the control method of the system comprises the following steps:

(1) in the running process of the vehicle, the vehicle-mounted distance and speed measuring sensing equipment transmits the real-time distance S between the vehicle and a dangerous target to the grading early warning control system, and transmits the motion state information such as the speed, the acceleration and the like of the dangerous target vehicle to the safe distance calculation model; the road surface information estimation model transmits the minimum road surface adhesion coefficient mu in the four wheels to a safe distance calculation model, and transmits the road surface gradient of the vehicle to a vehicle inverse longitudinal dynamics calculation model;

(2) the safe distance calculation model divides the braking intensity of the AEB system into a light braking mode and a full braking mode, and determines the target braking deceleration a of the two-stage braking intensity of the AEB system according to the minimum road adhesion coefficient mu in the four wheels_expAnd according to the dangerous target motion state information and the real-time longitudinal speed v of the vehicle measured by the vehicle speed sensor₀Determining three control signal trigger thresholds for the AEB system: sensory warning safety distance threshold S_wSafety distance threshold S for light braking_dSafety distance threshold S for full force braking_b；

(3) The grading early warning control system compares the real-time distance S between the vehicle and the dangerous target with early warning threshold values S at all levels_w、S_d、S_bIs compared, and whether the driver has active deceleration operation or not is combined, and the AEB early warning control signal, namely the target longitudinal deceleration a of the vehicle is analyzed and generated_exp: if S > S_wIf the distance between the vehicle and the front vehicle is within the safe range, the system judges that the vehicle and the front vehicle are not actuated; if S_w＜S＜S_dIf the driver is in the deceleration state, the system can perform visual touch early warning on the driver to remind the driver to perform deceleration operation; if S_b＜S＜S_dAnd judging that the driver does not perform deceleration operation at the moment according to the signal of the brake pedal stroke sensor, controlling the vehicle to perform mild braking by the system, decelerating the vehicle and performing sensory warning on the driver, wherein the target longitudinal deceleration a is at the moment_expFor light braking deceleration a_1maxThe mild braking deceleration a can effectively decelerate the vehicle and provide the driver with a tactile early warning, and simultaneously ensure that the comfort of passengers is not greatly influenced_1maxIt is preferably 0.25. mu.g to 0.35. mu.g; if the driver has deceleration operation at the moment, the AEB system can carry out sensory early warning on the driver until the driver keeps the distance between the vehicles at the safe distanceSeparating; if S is less than S_bThe system can control the vehicle to carry out full-force braking so that the vehicle is in the preset minimum safe vehicle distance range S at the fastest speed₀Internal stop or reaching the same speed as the dangerous target, minimum safe vehicle distance range S₀2-3 m, at which the target longitudinal deceleration a_expFor full-force braking deceleration a_2maxWhen full-force braking is triggered, full-force braking deceleration a is used to quickly decelerate the vehicle and prevent the wheels from locking up_2maxIt is preferably 0.75. mu.g to 0.85. mu.g;

(4) target longitudinal deceleration a_expThe target braking force F required to be provided by the vehicle braking system at the moment is finally obtained by transmitting the target braking force F to the vehicle inverse longitudinal dynamics calculation model, calculating the gradient i of the road surface where the vehicle is located by the road surface information estimation model at the same time, transmitting the gradient i to the vehicle inverse longitudinal dynamics calculation model, calculating by the formula (1) when the vehicle ascends the slope and calculating by the formula (2) when the vehicle descends the slope, and finally obtaining the target braking force F required to be provided by the vehicle braking_b：

F_b＝ma_exp-F_f-F_w-Gi (1)

F_b＝ma_exp+Gi-F_f-F_w(2)

Where m is the vehicle mass, G is the vehicle gravity, F_fTo rolling resistance, F_wIs the air resistance;

(5) target braking force F_bThe regenerative braking force F which can be provided at the moment is judged by the regenerative braking system according to the speed of the vehicle, the working state of systems such as a vehicle power motor, a storage battery and the like at the moment_brThe hydraulic braking force and regenerative braking force distribution calculation module calculates a target hydraulic braking force F at the time_bh＝F_b-F_br；

(6) Target hydraulic braking force F_bhThe target hydraulic pressure P of each brake wheel cylinder is obtained by the formula (3) after the target hydraulic pressure P is transmitted to the inverse model of the hydraulic brake system for calculation_exp：

Wherein r is_r0For wheel rolling radius, [ BEF]_fFor the front wheel brake braking efficiency factor, [ BEF]_rThe braking efficiency factors of the rear wheel brake are all conventional structural parameters of the vehicle;

(7) target hydraulic pressure P of each brake wheel cylinder_expTransmitting the pressure to an ESC and Booster active pressurization hydraulic pressure distribution module, and judging the active pressurization mode of the hydraulic braking system at the moment: if P_expIn the active pressurization limit range of the ESC system, the ESC system carries out active pressurization control on the pressure build-up of a brake wheel cylinder; if P_expIf the pressure exceeds the active pressure increase limit of the ESC system, the Booster builds pressure in a brake wheel cylinder to carry out active pressure increase control, so that the vehicle is decelerated to a target speed;

the above processes are continuously carried out in the whole triggering process of the AEB system, and the information such as the real-time distance between the vehicle and the dangerous target, the early warning threshold values of all levels, the motion states of the vehicle and the dangerous target and the like is continuously updated and adjusted until the vehicle is decelerated to the target speed or stops and keeps a preset safe distance between the vehicle and the dangerous target.

The invention has the beneficial effects that:

1. by adopting the safe distance calculation model and the grading early warning control system, according to the motion states of the vehicle and the front vehicle, the sensory early warning such as sound sense or vision can be carried out on the driver, the light braking or the full braking can be applied to the vehicle, the comfort when the AEB system is triggered is improved, and the potential safety hazards such as rear-end collision and the like caused by sudden and large deceleration of the vehicle are reduced.

2. By designing a road surface information estimation algorithm, information such as the gradient and the adhesion coefficient of a current running road and a front road of a vehicle is estimated, and sensors such as a camera are used for pre-estimating road surface information of a next stage, so that the practicability of the road surface information estimation algorithm is improved.

3. The method can perform online adjustment of control parameters according to related road information obtained by a road information estimation algorithm, enhance the adaptation degree of the AEB system to different road conditions, enable the AEB system to provide maximized safety performance for vehicles on any road, and simultaneously avoid the situation that the AEB system triggers too early to bring distrust to drivers.

4. For an electric automobile with a regenerative braking function, a regenerative braking system and a hydraulic braking system are coordinated to work, when the AEB system requires lower braking strength, the regenerative braking force and the hydraulic braking force are reasonably distributed, and the regenerative braking force of the automobile is exerted to the maximum extent, so that the braking energy is fully recovered, and the endurance mileage is improved.

5. When the AEB system requires lower braking strength and braking force can be provided by a regenerative braking system and the ESC system, the electronic mechanical brake Booster is not involved in braking, so that a brake pedal does not move autonomously, and the driving safety and comfort are improved; when the AEB system requires that the braking intensity is increased to exceed the sum of the active boosting limit of the ESC system and the maximum braking force which can be provided by the vehicle regenerative braking system at the moment, the vehicle is stopped as soon as possible or decelerated to the ideal speed by the intervention of the electronic mechanical brake Booster with stronger active boosting capacity.

Drawings

FIG. 1 is a schematic diagram of the overall architecture of the present invention;

FIG. 2 is a schematic diagram of a road information estimation model algorithm of the present invention;

FIG. 3 is a logic diagram of a hierarchical early warning control system according to the present invention;

FIG. 4 is a schematic diagram of the ESC and Booster active Booster hydraulic pressure distribution of the present invention;

FIG. 5 is a schematic diagram of a three-degree-of-freedom four-wheel vehicle model in the road adhesion coefficient estimation algorithm of the present invention;

FIG. 6 is a schematic flow chart of filtering estimation in the road adhesion coefficient estimation algorithm of the present invention;

FIG. 7 is a graph of a verification result of the road adhesion coefficient estimation algorithm of the present invention;

FIG. 8 is a schematic diagram illustrating a hierarchical braking process of the vehicle in the safe distance calculation model according to the present invention;

FIG. 9 is a schematic diagram illustrating a full braking process of the vehicle in the safe distance calculation model according to the present invention;

FIG. 10 is a graph of a road surface gradient estimation algorithm verification result based on a closed-loop full-dimensional state observer in the road surface information estimation algorithm of the present invention.

Detailed Description

Referring to fig. 1-10, the present invention provides an automatic-adjusting graded automatic emergency braking control system for electric vehicle, comprising: the system comprises vehicle-mounted distance and speed measuring sensing equipment, a graded early warning control system, a safe distance calculation model, a vehicle inverse longitudinal dynamics calculation model, a hydraulic braking force and regenerative braking force distribution calculation module, a hydraulic braking system inverse model, an ESC and Booster active pressurization hydraulic pressure distribution module and a road surface information estimation model, wherein the control method of the system comprises the following steps:

(1) referring to the attached figure 2, in the driving process of a vehicle, the vehicle-mounted distance and speed measuring sensing equipment transmits the real-time distance S between the vehicle and a dangerous target to a grading early warning control system, and transmits the motion state information such as the speed, the acceleration and the like of the dangerous target vehicle to a safe distance calculation model; the road surface information estimation model transmits the minimum road surface adhesion coefficient mu in the four wheels to a safe distance calculation model, and transmits the road surface gradient of the vehicle to a vehicle inverse longitudinal dynamics calculation model;

(2) the safe distance calculation model divides the braking intensity of the AEB system into a light braking mode and a full braking mode, and determines the target braking deceleration a of the two-stage braking intensity of the AEB system according to the minimum road adhesion coefficient mu in the four wheels_expAnd according to the dangerous target motion state information and the real-time longitudinal speed v of the vehicle measured by the vehicle speed sensor₀Determining three control signal trigger thresholds for the AEB system: sensory warning safety distance threshold S_wSafety distance threshold S for light braking_dSafety distance threshold S for full force braking_b；

(3) Referring to fig. 3, the hierarchical early warning control system compares the real-time distance S between the vehicle and the dangerous target with early warning thresholds S at different levels_w、S_d、S_bIs compared, and whether the driver has active deceleration operation or not is combined, and the AEB early warning control signal, namely the target longitudinal deceleration a of the vehicle is analyzed and generated_exp: if S > S_wIf the distance between the vehicle and the front vehicle is within the safe range, the system judges that the vehicle and the front vehicle are not actuated; if S_w＜S＜S_dIf the driver is in the low speed state, the system can carry out sensory early warning on the driver to remind the driver to carry out deceleration operation; if S_b＜S＜S_dAnd judging that the driver does not perform deceleration operation at the moment according to the signal of the brake pedal stroke sensor, controlling the vehicle to perform mild braking by the system, decelerating the vehicle and performing sensory warning on the driver, wherein the target longitudinal deceleration a is at the moment_expFor light braking deceleration a_1max(ii) a If the driver has deceleration operation at the moment, the AEB system can carry out sensory early warning on the driver until the driver keeps the distance between the vehicles at a safe distance; if S is less than S_bThe system can control the vehicle to carry out full-force braking, so that the vehicle can stop in a preset minimum safe vehicle distance range at the fastest speed or reach the speed same as the dangerous target, wherein the minimum safe vehicle distance range S₀The value is 2-3 m, and the target longitudinal deceleration a is carried out at the moment_expFor full-force braking deceleration a_2max；

(4) The grading early warning control system decelerates the target longitudinally a_expThe vehicle inverse longitudinal dynamics calculation model is based on a vehicle running resistance equation:

F＝F_f+F_w+F_i+F_j+F_b

wherein F is the longitudinal drag experienced by the vehicle, F_fTo rolling resistance, F_wAs air resistance, F_iIs the resistance of the slope, F_jFor acceleration resistance, F_bThe braking force generated by the vehicle braking system.

The vehicle inverse longitudinal dynamics calculation model receives an AEB early warning signal, namely a target longitudinal deceleration, transmitted by the hierarchical early warning control system, and when the hierarchical early warning control system requires to decelerate the vehicle, the driving resistance F in the formula is as follows:

F＝ma_exp

where m is the vehicle mass.

Since the vehicle is subjected to a deceleration operation, the acceleration resistance:

F_j＝0

meanwhile, the vehicle inverse longitudinal dynamics calculation model receives the road gradient i calculated by the road information estimation model, and if the vehicle needs to perform emergency braking on a slope, the influence of the gradient on the braking process must be considered:

①, if the vehicle is ascending, the slope resistance provides a part of the deceleration of the vehicle, and the slope resistance:

F_i＝Gi

wherein G is the vehicle gravity.

In combination, to obtain the desired longitudinal deceleration a of the vehicle_expThe braking system needs to provide the following braking force:

F_b＝ma_exp-F_f-F_w-Gi (1)

② if the vehicle is moving downhill, the component of the vehicle gravity along the slope will provide the vehicle with an acceleration downward along the slope, and it must be considered that this acceleration is offset by increasing the braking force, and the braking system needs to provide the following braking force:

F_b＝ma_exp+Gi-F_f-F_w(2)

therefore, the vehicle inverse longitudinal dynamics calculation model can calculate the longitudinal deceleration a according to the vehicle target transmitted by the grading early warning control system_expAnd judging the braking force F which should be provided by the braking system at the moment by combining the road surface gradient i estimated by the road surface information estimation model_bThe AEB system is prevented from being triggered prematurely or the vehicle is braked and stopped prematurely on an uphill road surface, so that the driver is not trusted; the danger of collision caused by insufficient braking due to too late intervention of an AEB system of a vehicle on a downhill road is avoided.

(5) Target braking force F_bThe regenerative braking force F which can be provided at the moment is judged by the regenerative braking system according to the speed of the vehicle, the working state of systems such as a vehicle power motor, a storage battery and the like at the moment_brThe hydraulic braking force and regenerative braking force distribution calculation module calculatesTarget hydraulic braking force F at this time_bh＝F_b-F_br(ii) a The hydraulic braking force and regenerative braking force distribution calculation module can fully exert the regenerative function braking capability of the electric vehicle to recover the braking energy, and simultaneously meet the braking deceleration requirement of the vehicle.

(6) Target hydraulic braking force F_bhAnd (3) transmitting the data to the hydraulic braking system inverse model for calculation, wherein the data is represented by the formula:

obtaining the target hydraulic pressure P of each brake wheel cylinder_exp：

Wherein r is_r0For wheel rolling radius, [ BEF]_fFor the front wheel brake braking efficiency factor, [ BEF]_rThe braking efficiency factors of the rear wheel brake are all conventional structural parameters of the vehicle;

(7) referring to fig. 4, target hydraulic pressure P of each brake cylinder_expTransmitting the pressure to an ESC and Booster active pressurization hydraulic pressure distribution module, and judging the active pressurization mode of the hydraulic braking system at the moment: if P_expIn the active pressurization limit range of the ESC system, the ESC system carries out active pressurization control on the pressure build-up of a brake wheel cylinder; if P_expIf the pressure exceeds the active pressure increase limit of the ESC system, the Booster builds pressure in a brake wheel cylinder to carry out active pressure increase control, so that the vehicle is decelerated to a target speed;

the ESC and Booster active pressurization hydraulic pressure distribution module can fully utilize the advantages of an ESC system and the Booster active pressurization: when the braking pressure requirement is not high, an ESC system with relatively low response time and precision can be utilized, the pressure requirement is met, and meanwhile, for a vehicle carrying a non-decoupling braking system, the discomfort brought to a driver by automatic downward movement of a brake pedal caused by boost active pressurization can be avoided, and the potential safety hazard is reduced; when the braking pressure requirement is high, Booster with high response speed, high pressure control precision and high building upper limit is adopted for active braking, and the vehicle can achieve the expected braking deceleration by full force.

The above processes are continuously carried out in the whole triggering process of the AEB system, and the information such as the real-time distance between the vehicle and the dangerous target, the early warning threshold values of all levels, the motion states of the vehicle and the dangerous target and the like is continuously updated and adjusted until the vehicle is decelerated to the target speed or stops and keeps a preset safe distance between the vehicle and the dangerous target.

In the step (1), the method for obtaining the minimum road adhesion coefficient mu by the road adhesion coefficient estimation model based on filtering is as follows:

the Dugoff tire model is first built to obtain normalized tire forces, and represents the longitudinal and lateral forces on the tire as:

in the formula F_xIs a tire longitudinal force, F_yFor lateral forces of the tire, F_zIs the tire normal force, λ is the longitudinal slip ratio, C_yFor the cornering stiffness of the tyre, C_xFor tyre longitudinal rigidity, α is tyre slip angle, mu is road surface adhesion coefficient, epsilon is tyre influence coefficient, which is a parameter related to tyre structure and material and used for correcting the influence of vehicle slip speed on tyre force, L is boundary value and used for describing the non-linear characteristic of tyre force brought by wheel slip, and for designing road surface adhesion coefficient estimation algorithm conveniently, the Dugoff tyre model is simplified into the following normalization formFormula (II):

in the formula

And

the normalized tire forces in the longitudinal direction and the lateral direction are respectively independent of the adhesion coefficient mu, so that a coefficient matrix of a system state space expression is conveniently determined in the road adhesion coefficient estimation algorithm based on filtering;

after the normalized Dugoff tire model is established, the normalized tire force in the model needs to be calculated, namely the vertical load F of each wheel is calculated respectively_Zfl、F_Zfr、F_Zrl、F_ZrrLongitudinal slip ratio λ_fl、λ_fr、λ_rl、λ_rrTire slip angle α_fl，α_fr，α_rl，α_rrThe corner marks fl, fr, rl, rr represent the left front wheel, the right front wheel, the left rear wheel, and the right rear wheel of the vehicle, respectively;

the vertical load calculation formula of each wheel is as follows:

the formula for calculating the slip angle of each wheel is as follows:

calculation formula of each wheel slip ratio:

velocity v of each wheel grounding point in the above formula_fl、v_fr、v_rl、v_rrThe calculation formula is as follows:

the meaning of each parameter in the above formulas is β is the vehicle mass center slip angle,

v_cogis the speed of the center of mass of the vehicle,

m is the total vehicle mass, a is the horizontal distance between the center of the front wheel and the vehicle mass center, b is the horizontal distance between the center of the rear wheel and the vehicle mass center, l is the vehicle wheel base, h_gIs the height of the center of mass of the vehicle, a_xFor longitudinal acceleration of the vehicle, a_yFor lateral acceleration of the vehicle, T_fIs a track of two front wheels, T_rIs a two rear wheel track v_xLongitudinal speed, v, of the entire vehicle_yLateral speed of the entire vehicle, R_ωAs wheel rolling speed, ω_fl、ω_fr、ω_rl、ω_rrRolling angular velocities of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel are respectively;

the longitudinal normalized tire force for the four wheels can thus be expressed as:

the lateral normalized tire force for the four wheels can be expressed as:

parameters required in the calculation process of the normalized tire force corrected by combining the Dugoff tire model comprise vehicle structure parameters and kinematic parameters, the structure parameters can be directly measured, and the kinematic parameters can be measured by a related vehicle-mounted sensor;

in order to write a system state equation and an observation equation in a filtering-based road adhesion coefficient estimation algorithm in a row, a three-degree-of-freedom four-wheel vehicle model of a vehicle is required to be established to describe the longitudinal, lateral and yaw motions of the vehicle;

referring to fig. 5, the equations of motion for the vehicle in the three directions, longitudinal, lateral and yaw, are:

in the formula, a_xFor longitudinal acceleration of the vehicle, a_yIn order to provide for a lateral acceleration of the vehicle,

is the yaw angular acceleration of the vehicle, delta is the two front wheel corners, m is the total vehicle mass, mu_fl、μ_fr、μ_rl、μ_rrRespectively, the road surface adhesion coefficients of a left front wheel, a right front wheel, a left rear wheel and a right rear wheel, a is the horizontal distance between the center of the front wheel and the mass center of the vehicle, b is the horizontal distance between the center of the rear wheel and the mass center of the vehicle, and T is the horizontal distance between the center of the rear wheel and the mass center of the vehicle_fIs a track of two front wheels, T_rTwo rear wheel track, I_zYaw moment of inertia of the vehicle;

the method comprises the following steps that related sensors on a vehicle measure vehicle kinematic parameters required by an algorithm and transmit the vehicle kinematic parameters to a Dugoff tire model, the Dugoff tire model calculates longitudinal and lateral normalized tire forces and transmits the longitudinal and lateral normalized tire forces to a vehicle three-degree-of-freedom four-wheel model, so that three dynamic equations in the longitudinal direction, the lateral direction and the yaw direction are obtained, and a system state equation and an observation equation of a road adhesion coefficient filter estimator can be obtained based on the three dynamic equations to carry out subsequent adhesion coefficient estimation;

the system selected by the road adhesion coefficient estimation algorithm based on filtering is a nonlinear system, and the state equation and the observation equation of the system are respectively as follows:

the state equation is as follows:

the observation equation:

y(t)＝h(x(t),u(t),v(t))

the random variables w (t), v (t) are respectively process noise and measurement noise, Gaussian white noise which are independent from each other and have a zero mean value is taken in the filtering-based road adhesion coefficient estimation algorithm, and the probability distribution is as follows:

p(w)～N(0,Q)p(v)～N(0,R)

and setting their covariance matrices as Q and R, respectively, namely:

Q＝cov[w(t),w(τ)]

R＝cov[v(t),v(τ)]

referring to fig. 6, the filtering-based road adhesion coefficient estimation algorithm is implemented as follows:

(1) determining a system state equation and an observation equation:

determining according to the motion equations of the longitudinal direction, the lateral direction and the yaw direction of the three-degree-of-freedom four-wheel vehicle model:

the state variables are as follows: x (t) ═ μ_flμ_frμ_rlμ_rr]^T，

Observation variables: y (t) ═ a_xa_yr]^T，

And (3) control input: u (t) ═ δ,

writing a state equation and an observation equation into the three-degree-of-freedom four-wheel vehicle model according to expressions and variables of motion equations in the longitudinal direction, the lateral direction and the yaw direction of the three-degree-of-freedom four-wheel vehicle model:

the state equation is as follows:

the observation equation:

wherein:

(2) the estimator assigns an initial value:

measuring an initial value in a recursion process, wherein a noise covariance matrix R is a 3-by-3 unit matrix, and a process noise covariance matrix Q is a 4-by-4 unit matrix;

(3) estimating the road adhesion coefficient by using a filtering algorithm:

①, the filter is initialized, the iteration step number k is 0, and the initial mean value of the state variable x

And the covariance P (0) are:

② calculating by using unscented transformation to obtain 2n +1 points, making the mean and covariance of the 2n +1 points equal to those of the original state distribution, and referring the 2n +1 points as Sigma points, n as state dimension, according to the above state equation, taking n as 4, then obtaining 9 Sigma point sets, where the state dimension of each Sigma point set is consistent with that of state variable x, that is, 4-dimensional column vector, then the matrix x formed by the 9 Sigma point sets is 4 x 9 matrix, and setting χ_i(0) For each column vector, i.e. each Sigma point set, χ in the matrix χ at iteration step k ═ 0₀(0) Column one of Sigma matrix χ, and so on, the initial Sigma matrix is:

where λ is a proportionality parameter, λ ═ α²-1)·n＝4(α²-1), α for determining the Sigma point at the mean of the state variable

The nearby distribution range is a very small positive number 10^-4α ≤ 1, and α ═ 10^-3；

The above-mentioned calculation process of the initial Sigma point set when the iteration step number k is 0 is also applicable to the Sigma point set calculation of other iteration step numbers, except that the state variables used for the Sigma point set calculation at this time are allThe value and state variable covariance values being the value at time k, i.e. at time k

And p (kk), the calculation formula is:

③, proceeding to the next iteration, increasing the number k of iteration steps by 1, predicting the 9 Sigma point sets in the next step, substituting the Sigma point set obtained in the previous step in the step ② into the state equation (a) of the system to obtain a transformed Sigma point set:

χ(kk-1)＝f(χ(k-1),u(k-1))k＝1,2,…

④, calculating the next prediction mean value and the prediction variance of the system state variable according to the principle of unscented transformation, wherein the prediction mean value of the system state variable is obtained by the weighted summation of the prediction values of the Sigma point set:

in the formula W_i ^(m)The average weight of each Sigma point is specifically:

W₀ ^(m)＝λ/(4+λ)i＝0

W_i ^(m)＝1/[2(4+λ)]i＝1,2,…,8

the predicted variance of the system state variables is obtained by weighted summation of the predicted covariance of the Sigma point set:

middle x of the above formula_i(k | k-1) is the ith column of the matrix χ (k | k-1), i ═ 0,1, …, 8; w_i ^(c)The covariance weight of each Sigma point is specifically:

W₀ ^(c)＝λ/(n+λ)+(1-α²+β)i＝0

W_i ^(c)＝1/[2(n+λ)]i＝1,2,…,8

wherein β is used to merge prior states relating to the distribution of state variables x, which obey a gaussian distribution in the filter-based road adhesion coefficient estimation algorithm, typically β -2;

⑤ generating a new Sigma point set by using the unscented transformation again according to the next predicted mean and the next predicted variance of the system state variables obtained in step ④, the calculation process is the same as that of the initial Sigma point set in step ②:

⑥ substituting the new Sigma point set obtained in step ⑤ into observation equation (b) to obtain the next predicted value of the observed quantity of the Sigma point set:

u (k-1) in the formula is system input at the moment when the iteration step number is k-1, namely the front wheel rotation angle delta at the moment can be measured by the existing vehicle-mounted rotation angle sensor;

according to the observation equation (b), the output matrix of the system is a matrix of 3 times 4, and the Sigma matrix chi (kk-1) is a matrix of 4 times 9, so that the next predicted value of the observed quantity of the Sigma point set is calculated through the observation equation (b)

A matrix of 3 by 9;

⑦ predicted value from the observed value of Sigma Point set obtained in step ⑥

Obtaining a predicted mean value of the system observation variable through weighted summation:

in the formula

Matrix of next prediction values of observed quantities of Sigma point set

I ═ 0,1, …, 8;

⑧ predicted value from the observed value of Sigma Point set obtained in step ⑥

Calculating an updated observation covariance matrix and a state variable and output variable cross-correlation matrix by weighted summation:

the observed covariance matrix is:

the state variable and output variable cross-correlation matrix is:

⑨ calculating the updated filtered feedback gain matrix of the system:

calculating the updated state variable mean matrix of the system:

matrix array

The (1,1) element, (2,1) element, (3,1) element, and (4,1) element are the filter estimation values of the left front wheel adhesion coefficient, the right front wheel adhesion coefficient, the left rear wheel adhesion coefficient, and the right rear wheel adhesion coefficient when the iteration step number is k, respectively;

wherein y (k) is the actual observed variable value at the moment when the iteration step number is k, i.e. the longitudinal acceleration a at the moment when the iteration step number is k_xLateral acceleration a_yAnd yaw angular acceleration

The three vehicle kinematic parameters can be measured in real time by a relevant acceleration sensor assembled on the vehicle and then transmitted to the road adhesion coefficient estimation algorithm based on filtering;

calculating the updated state variable covariance value of the system:

P(k|k)＝P(k|k-1)-K(k)P_yyK^T(k)

mean value of state variable after system update

And the state variable covariance value P (k | k) returns to step ② to generate the next set of Sigma point sets, and the calculation of the next iteration step is started, wherein the process is repeated continuously until all the iteration steps are completed, and finally the road adhesion coefficients mu of the four wheels are obtained;

the iteration step number k of the filtering-based road adhesion coefficient estimation algorithm depends on the sampling step length set by the algorithm, and the selection of the sampling step length directly influences the final convergence degree and the final convergence speed of the algorithm, namely the estimation speed and the accuracy of the road adhesion coefficients of the four wheels, so that the sampling step length of the algorithm needs to be specifically adjusted according to related parameters of different vehicles to enable the algorithm to obtain the best performance.

With reference to fig. 7, the road adhesion coefficient estimation algorithm based on filtering is verified, and it can be seen from the drawing that the road adhesion coefficients of the wheels can all be converged within 1s, and are basically consistent with the actual reference values of the road adhesion coefficients, and the error is extremely small, which shows that the road adhesion coefficient estimation algorithm based on filtering of the invention has good performance, the road adhesion coefficient estimation speed and the estimation precision are extremely high, and the requirements of the vehicle-level algorithm are met.

The road adhesion coefficient estimation algorithm based on filtering considers the situation that a vehicle runs on the road surface with different four wheel adhesion coefficients such as an open road surface, after the road adhesion coefficient estimation value of each wheel is obtained, the road adhesion coefficient estimation algorithm based on filtering compares the four values to obtain the minimum value of the four wheel adhesion coefficients, and feeds the minimum value mu back to the safe distance calculation model so as to prevent the collision danger caused by insufficient vehicle braking deceleration due to the fact that sufficient braking force cannot be generated between a low-adhesion-coefficient tire and the road surface.

In step (2), the target braking deceleration a of the light braking of the AEB system_1max0.25 to 0.35 [ mu ] g, target braking deceleration a of full-force braking_2max0.75-0.85 mug; the three-level early warning safety distance threshold value calculation method of the AEB system comprises the following steps:

referring to fig. 8, from the perspective of vehicle braking, the hierarchical braking process of the vehicle is divided into six stages:

the first stage is as follows: the duration of the period t from the initiation of the mild braking signal by the AEB system to the initiation of the mild braking deceleration of the vehicle by the active braking system₁₁Determined by the response lag time of the active braking system of the vehicle, can be obtained through experimental tests. At this stage, since the brake system is not building up brake pressure, the vehicle deceleration is 0, and the displacement of the host vehicle is:

S₁₁＝v₀t₁₁

in the formula: v. of₀The initial speed of the vehicle;

and a second stage: the stage takes the start of pressurization of the active braking system as a starting point until the braking pressure reaches the target hydraulic pressure of mild braking, and the pressurization time is t₁₂(ii) a In this process, the vehicle braking deceleration a increases with the increase of the brake pressure₁Vehicle speed v₁₂And a displacement S₁₂The expressions are respectively as follows:

v₁₂＝v₀-∫a₁·dt

in the formula: a is_1maxTarget braking deceleration for a light braking stage;

and a third stage: this phase is a mild braking phase with the AEB system stable, with the vehicle remaining a_1maxIs constant, the duration t of the process₁₃The speed v of the vehicle is set by an AEB system and can be 1-2 s in the process₁₃And a displacement S₁₃The changes are as follows:

v₁₃＝v₀-0.5a_1maxt₁₂-a_1maxt

a fourth stage: the phase from the initiation of the emergency full force braking signal by the AEB system to the initiation of the boost by the active braking system, during which the vehicle remains a_1maxIs constant in braking deceleration, this phase responding to the lag duration t₂₁＝t₁₁(ii) a Vehicle speed v at this stage₂₁And a displacement S₂₁Respectively as follows:

v₂₁＝v₀-a_1max(0.5t₁₂+t₁₃)-a_1maxt

the fifth stage: the stage is a boosting stage of the hydraulic braking system, the active braking pressure is increased from a mild braking target value to a full force braking target value, and the boosting time t₂₂(ii) a At this stage, the vehicle deceleratesDegree a₂Vehicle speed v₂₂And a displacement S₂₂The expressions are respectively as follows:

v₂₂＝v₀-a_1max(0.5t₁₂+t₁₃+t₂₁)-∫a₂dt

in the formula: a is_2maxThe target braking deceleration is the full-force braking stage;

the sixth stage: a stable full-force braking stage, which is started from the time when the braking pressure of the vehicle reaches the target braking pressure, the time when the vehicle speed is reduced to 0 or the time when the vehicle speed is reduced to the dangerous target vehicle speed v_tUntil the end; in this phase, the vehicle initial speed v₂Duration t₂₃Real-time vehicle speed v₂₃And a braking distance S₂₃The expressions are respectively as follows:

v₂＝v₀-a_1max(0.5t₁₂+t₁₃+t₂₁+0.5t₂₂)-0.5a_2maxt₂₂

v₂₃＝v₂-a_2max·t

the calculation process of the graded braking of the vehicle is finished;

because the dangerous target vehicle in front has the possibility of emergency braking, under the working condition, the vehicle takes the distance between two vehicles which are subjected to emergency braking simultaneously and do not collide as the safe distance threshold value of light braking, the full-force braking process of the vehicle active braking system is calculated:

the process is divided into three stages:

the first stage is as follows: a hysteresis phase for the response of the active braking system, during which the vehicle travels a distance S₁Comprises the following steps:

S₁＝v₀·t₁

and a second stage: an active braking pressure establishing stage in which the active braking pressure is linearly increased, a braking deceleration a of the vehicle and a vehicle speed v of the vehicle₁And the vehicle displacement S₂Are respectively:

v₁＝v₀-∫adt

in the formula: t is t₂Time of pressure build-up for active braking system, a_2maxA target braking deceleration for the host vehicle;

and a third stage: in which the vehicle is decelerated uniformly until it stops, and in which the initial speed v of the vehicle₃₀Real-time vehicle speed v₃Duration t₃And a braking distance S₃Respectively as follows:

v₃＝v₃₀-a_2max·t

next, AEB three early warning levels are carried out to obtain a safety distance threshold value S_wS_dS_bThe calculation of (2):

① full force braking threshold S_b：

The current dangerous target vehicle has the speed v known from the graded braking process of the vehicle_tThe vehicle keeps running at a constant speed, and when the vehicle is braked and decelerated to the front dangerous target vehicle speed by full force, the braking distance S of the vehicle_{h1_m}Comprises the following steps:

S_{h1_m}＝S₂₁+S₂₂+S₂₃

in this process, the distance S traveled by the front dangerous target vehicle_t1Comprises the following steps:

S_t1＝v_t(t₂₁+t₂₂+t₂₃)

the target distance which can be kept with the front vehicle after the vehicle finishes the automatic emergency braking is S₀Then the AEB system triggers the safe distance threshold S of full force braking_bComprises the following steps:

S_b＝S_{h1_m}-S_t1+S₀

② light braking safety distance threshold S_d：

When the front dangerous target vehicle suddenly brakes at the maximum deceleration, the braking distance S is_t2Comprises the following steps:

the vehicle also adopts a full-force braking model to calculate the braking distance S_{h2_m}Comprises the following steps:

S_{h2_m}＝S₁+S₂+S₃

the AEB system light braking safety distance threshold value S_dComprises the following steps:

S_d＝S_{h2_m}-S_t2+S₀

③ visual and auditory early warning safety distance threshold S_w：

S_w＝S_d+t_w·v₀

In the formula t_wThe duration of the early warning for the sound sensation and the vision can be selected as1～1.5s。

It can be seen from the above AEB early warning level safety threshold that the mild braking deceleration of the AEB system in the safety distance calculation model is 0.3 μ g, the braking deceleration of the full-force braking is 0.8 μ g, and μ is the minimum value of the road surface adhesion coefficients of the four wheels of the vehicle at that time, which is calculated by the road surface information estimation model, so that the deceleration values of the mild braking and the full-force braking can be changed online in real time according to the actual minimum road surface adhesion coefficient μ, and the collision danger caused by the fact that the road surface cannot provide the preset braking deceleration when the AEB system performs automatic braking is prevented. Meanwhile, the safe distance calculation model can adjust early warning threshold values at all levels according to the road adhesion coefficient at the moment, and the danger caused by too-late triggering of the AEB system is prevented.

In the step (4), the method for estimating the gradient i of the road surface where the vehicle is located by the road surface information estimation model based on the road surface gradient estimation algorithm of the closed-loop full-dimensional state observer is as follows:

the related sensors of the vehicle transmit vehicle parameters required by road gradient estimation to a road information estimation model, and the vehicle running equation is F_t＝F_f+F_w+F_i+F_j；

Vehicle driving force F_tComprises the following steps:

where r is the rolling radius of the wheel, T_eAs vehicle engine torque, i_gFor the transmission ratio of the vehicle transmission, i₀As a vehicle final drive, η_tThe overall efficiency of the vehicle driveline;

air resistance F of vehicle_wComprises the following steps:

wherein C is_DIs the coefficient of air resistance, A is the frontal area, ρ is the air density, v_xIs the vehicle longitudinal speed;

vehicle acceleratorFast resistance F_jComprises the following steps:

F_j＝δMa_x

wherein δ is a vehicle rotating mass conversion coefficient; m is the total vehicle mass, a_xIs the vehicle longitudinal acceleration;

rolling resistance F of vehicle_fComprises the following steps:

F_f＝Mgfcosα≈Mgf

wherein f is the vehicle rolling resistance coefficient;

gradient resistance F of vehicle_iComprises the following steps:

F_i＝Mgsinα≈Mgi

the longitudinal dynamic equation of the vehicle on the slope can be written as:

the vehicle-related parameters involved in the calculation process are all parameters which are easily obtained in the vehicle design and manufacturing process. In order to apply the state observer, the longitudinal dynamic equation is linearized, and the vehicle driving force F is processed_tAir resistance F_wAnd rolling resistance F_fViewed as a resultant force F_inputAnd (3) inputting a longitudinal kinetic equation as a system, and simplifying the longitudinal kinetic equation into:

δMa＝F_input-Mgi

the above formula is a system state equation, and a state space expression of the system can be written based on the system state equation; at vehicle longitudinal speed v_xAnd gradient i as system state variable, with resultant force F_inputFor system input variables, the state space expression based on the automobile longitudinal dynamics equation is as follows:

z＝Cx

wherein:

C＝[1 0]；

the observability of the system is verified as follows, and the observability matrix of the system is:

observability matrix Q_BThe full rank indicates that the system can observe, so the road surface slope observer is designed as follows:

wherein

Is an observation vector of the observer,

h is the feedback gain matrix of the observer; e is an error vector, i.e.

Will be provided with

And

subtraction can give:

is provided with

J, the above formula is changed to

Is converted into a first-order linear differential equation about j, and the initial time is set as t₀The differential equation is solved as:

to guarantee observer stability, it should be guaranteed that the observer error vector e is equal to 0 when the time goes to infinity, i.e.:

as long as the characteristic root λ of (a-HC) has a negative real part, the error vector will decay exponentially to 0, and the decay rate is determined by the characteristic root of (a-HC); setting feedback gain matrix of observer

The characteristic equation of (A-HC) is:

the expected pole of the full-dimensional observer, i.e. the (A-HC) characteristic root with a negative real part, is preset to be lambda₁And λ₂，λ₁And λ₂Can be selected by the user of the system, as long as the system is ensured to have a negative real part, and the pole lambda is expected₁And λ₂The corresponding expected characteristic equation is:

(λ-λ₁)(λ-λ₂)＝λ²-(λ₁+λ₂)λ+λ₁λ₂(5)

let the elements of the feedback gain matrix H of the available observer with the same term coefficients for equations (4) and (5) be:

h₁＝-λ₁-λ₂

the characteristic value lambda can be ensured by calculating and obtaining the feedback gain matrix H according to the method₁And λ₂Has a negative real part, even if the observer is stable;full-dimensional state observer observation variable

Second row element of (1)

I.e. the estimated value i of the road gradient on which the vehicle is located at the moment.

Referring to the attached drawing 10, the road surface gradient estimation algorithm based on the closed-loop full-dimensional state observer is verified, and as can be seen from the attached drawing, the road surface gradient estimation value can be converged in a very short time, and is basically consistent with an actual reference value of the road surface gradient, and the error is very small.

Claims (3)

1. The utility model provides an automatic hierarchical automatic emergency brake control system of electric motor car of adjustment which characterized in that: the control system comprises: the system comprises vehicle-mounted distance and speed measuring sensing equipment, a graded early warning control system, a safe distance calculation model, a vehicle inverse longitudinal dynamics calculation model, a hydraulic braking force and regenerative braking force distribution calculation module, a hydraulic braking system inverse model, an ESC and Booster active pressurization hydraulic pressure distribution module and a road surface information estimation model, wherein the control method of the system comprises the following steps:

(1) in the running process of the vehicle, the vehicle-mounted distance and speed measuring sensing equipment transmits the real-time distance S between the vehicle and the dangerous target vehicle to the grading early warning control system, and transmits the motion state information of the dangerous target vehicle to the safe distance calculation model; the road surface information estimation model transmits the minimum road surface adhesion coefficient mu in the four wheels of the vehicle to the safe distance calculation model, and transmits the road surface gradient of the vehicle to the vehicle inverse longitudinal dynamics calculation model;

(2) the safe distance calculation model divides the braking intensity of the AEB system into a light braking mode and a full braking mode, and determines the target braking deceleration of the two-stage braking intensity of the AEB system according to the minimum road adhesion coefficient mu in the four wheelsa_expTarget braking deceleration a of light braking_expGet a_1max0.25 to 0.35 [ mu ] g, target braking deceleration a of full-force braking_expGet a_2max0.75-0.85 mug; and the real-time longitudinal speed v of the vehicle is measured by the vehicle speed sensor according to the motion state information of the dangerous target₀Determining three control signal trigger thresholds for the AEB system: sensory warning safety distance threshold S_wSafety distance threshold S for light braking_dSafety distance threshold S for full force braking_bThe three-level early warning safety distance threshold value calculation method of the AEB system comprises the following steps:

from the perspective of vehicle braking, the hierarchical braking process of the vehicle is divided into six stages:

the first stage is as follows: the duration of the period t from the initiation of the mild braking signal by the AEB system to the initiation of the mild braking deceleration of the vehicle by the active braking system₁₁The response lag time of the active braking system of the vehicle, at this stage, since the braking system is not building up brake pressure, the deceleration of the vehicle is 0, and the displacement of the host vehicle is:

S₁₁＝v₀t₁₁

in the formula: v. of₀The initial speed of the vehicle;

and a second stage: the stage takes the start of pressurization of the active braking system as a starting point until the braking pressure reaches the target hydraulic pressure of mild braking, and the pressurization time is t₁₂(ii) a In this process, the vehicle braking deceleration a increases with the increase of the brake pressure₁Vehicle speed v₁₂And a displacement S₁₂The expressions are respectively as follows:

v₁₂＝v₀-∫a₁·dt

vehicle braking decelerationA of degree₁In the formula: a is_1maxThe target braking deceleration in the mild braking stage is t, and the t represents an independent variable of time;

and a third stage: this phase is a mild braking phase with the AEB system stable, with the vehicle remaining a_1maxIs constant, the duration t of the process₁₃The speed v of the vehicle is set by an AEB system and can be 1-2 s in the process₁₃And a displacement S₁₃The changes are as follows:

v₁₃＝v₀-0.5a_1maxt₁₂-a_1maxt

vehicle speed v₁₃In the formula: t represents an argument of time;

a fourth stage: the phase from the initiation of the emergency full force braking signal by the AEB system to the initiation of the boost by the active braking system, during which the vehicle remains a_1maxIs constant in braking deceleration, this phase responding to the lag duration t₂₁＝t₁₁(ii) a Vehicle speed v at this stage₂₁And a displacement S₂₁Respectively as follows:

v₂₁＝v₀-a_1max(0.5t₁₂+t₁₃)-a_1maxt

vehicle speed v₂₁In the formula: t represents an argument of time;

the fifth stage: the stage is a boosting stage of the hydraulic braking system, the active braking pressure is increased from a mild braking target value to a full force braking target value, and the boosting time t₂₂(ii) a At this stage, the vehicle deceleration a₂Vehicle speed v₂₂And a displacement S₂₂The expressions are respectively as follows:

v₂₂＝v₀-a_1max(0.5t₁₂+t₁₃+t₂₁)-∫a₂dt

deceleration a of the vehicle₂In the formula: a is_2maxT represents the independent variable of time as the target braking deceleration in the full-force braking stage;

the sixth stage: a stable full-force braking stage, which is started from the time when the braking pressure of the vehicle reaches the target braking pressure, the time when the vehicle speed is reduced to 0 or the time when the vehicle speed is reduced to the dangerous target vehicle speed v_tUntil the end; in this phase, the vehicle initial speed v₂Duration t₂₃Real-time vehicle speed v₂₃And a braking distance S₂₃The expressions are respectively as follows:

v₂＝v₀-a_1max(0.5t₁₂+t₁₃+t₂₁+0.5t₂₂)-0.5a_2maxt₂₂

v₂₃＝v₂-a_2max·t

real-time vehicle speed v₂₃In the formula: t represents an argument of time;

the calculation process of the graded braking of the vehicle is finished;

because the dangerous target vehicle in front has the possibility of emergency braking, under the working condition, the vehicle takes the distance between two vehicles which are subjected to emergency braking simultaneously and do not collide as the safe distance threshold value of light braking, the full-force braking process of the vehicle active braking system is calculated:

the process is divided into three stages:

the first stage is as follows: a hysteresis phase for the response of the active braking system, during which the vehicle travels a distance S₁Comprises the following steps:

S₁＝v₀·t₁

and a second stage: an active braking pressure establishing stage in which the active braking pressure is linearly increased, a braking deceleration a of the vehicle and a vehicle speed v of the vehicle₁And the vehicle displacement S₂Are respectively:

v₁＝v₀-∫adt

the braking deceleration a of the vehicle is expressed by the formula: t is t₂Time of pressure build-up for active braking system, a_2maxT represents an independent variable of time for a target braking deceleration of the host vehicle;

and a third stage: in which the vehicle is decelerated uniformly until it stops, and in which the initial speed v of the vehicle₃₀Real-time vehicle speed v₃Duration t₃And a braking distance S₃Respectively as follows:

v₃＝v₃₀-a_2max·t

real-time vehicle speed v₃In the formula: t represents an argument of time;

next, AEB three early warning levels are carried out to obtain a safety distance threshold value S_w、S_d、S_bThe calculation of (2):

① full force braking threshold S_b：

The current dangerous target vehicle has the speed v known from the graded braking process of the vehicle_tThe vehicle keeps running at a constant speed, and when the vehicle is braked and decelerated to the front dangerous target vehicle speed by full force, the braking distance S of the vehicle_{h1_m}Comprises the following steps:

S_{h1_m}＝S₂₁+S₂₂+S₂₃

in this process, the distance S traveled by the front dangerous target vehicle_t1Comprises the following steps:

S_t1＝v_t(t₂₁+t₂₂+t₂₃)

the minimum safe vehicle distance range S is the target distance which can be kept between the vehicle and the front vehicle after the vehicle finishes the automatic emergency braking₀，S₀The distance is 2-3 m, the AEB system triggers the safe distance threshold value S of full-force braking_bComprises the following steps:

S_b＝S_{h1_m}-S_t1+S₀

② light braking safety distance threshold S_d：

When the front dangerous target vehicle suddenly brakes at the maximum deceleration, the braking distance S is_t2Comprises the following steps:

the vehicle also adopts a full-force braking model to calculate the braking distance S_{h2_m}Comprises the following steps:

S_{h2_m}＝S₁+S₂+S₃

the AEB system light braking safety distance threshold value S_dComprises the following steps:

S_d＝S_{h2_m}-S_t2+S₀

③ visual and auditory early warning safety distance threshold S_w：

S_w＝S_d+t_w·v₀

In the formula t_wThe duration of the early warning for the sound sensation and the vision can be 1-1.5 s;

(3) the grading early warning control system compares the real-time distance S between the vehicle and the dangerous target with early warning threshold values S at all levels_w、S_d、S_bIs compared, and whether the driver has active deceleration operation or not is combined, and the AEB early warning control signal, namely the target longitudinal deceleration a of the vehicle is analyzed and generated_exp: if S > S_wIf the distance between the vehicle and the front vehicle is within the safe range, the system judges that the vehicle and the front vehicle are not actuated; if S_w＜S＜S_dIf the driver is in the deceleration state, the system can perform visual touch early warning on the driver to remind the driver to perform deceleration operation; if S_b＜S＜S_dAnd judging that the driver does not perform deceleration operation at the moment according to the signal of the brake pedal stroke sensor, the system controls the vehicle to perform light braking, and the target longitudinal deceleration a is performed at the moment_expFor light braking deceleration a_1max(ii) a If the driver has deceleration operation at the moment, the AEB system can carry out sensory early warning on the driver until the driver keeps the distance between the vehicles at a safe distance; if S is less than S_bThe system can control the vehicle to carry out full-force braking so that the vehicle is in the preset minimum safe vehicle distance range S at the fastest speed₀Stopping inside or reaching the same speed as the dangerous target, at which time the target longitudinal deceleration a_expFor full-force braking deceleration a_2max；

(4) The grading early warning control system decelerates the target longitudinally a_expThe target braking force F required to be provided by the vehicle braking system at the moment is finally obtained by transmitting the target braking force F to the vehicle inverse longitudinal dynamics calculation model, calculating the gradient i of the road surface where the vehicle is located by the road surface information estimation model at the same time, transmitting the gradient i to the vehicle inverse longitudinal dynamics calculation model, calculating by the formula (1) when the vehicle ascends the slope and calculating by the formula (2) when the vehicle descends the slope, and finally obtaining the target braking force F required to be provided by the vehicle braking_b：

F_b＝ma_exp-F_f-F_w-Gi (1)

F_b＝ma_exp+Gi-F_f-F_w(2)

Where m is the vehicle mass, G is the vehicle gravity, F_fTo rolling resistance, F_wIs the air resistance;

(5) target braking force F_bThe distribution calculation module transmits the hydraulic braking force and the regenerative braking force to the regenerative braking system, and the regenerative braking system judges the regenerative braking force F which can be provided at the moment according to the working state of the vehicle at the moment_brThe hydraulic braking force and regenerative braking force distribution calculation module calculates a target hydraulic braking force F at the time_bh＝F_b-F_br；

(6) Target hydraulic braking force F_bhThe target hydraulic pressure P of each brake wheel cylinder is obtained by the formula (3) after the target hydraulic pressure P is transmitted to the inverse model of the hydraulic brake system for calculation_exp：

Wherein r is_r0For wheel rolling radius, [ BEF]_fFor the front wheel brake braking efficiency factor, [ BEF]_rThe braking efficiency factors of the rear wheel brake are all conventional structural parameters of the vehicle;

(7) target hydraulic pressure P of each brake wheel cylinder_expTransmitting the pressure to an ESC and Booster active pressurization hydraulic pressure distribution module, and judging the active pressurization mode of the hydraulic braking system at the moment: if P_expIn the active pressurization limit range of the ESC system, the ESC system carries out active pressurization control on the pressure build-up of a brake wheel cylinder; if P_expIf the pressure exceeds the active pressure increase limit of the ESC system, the Booster builds pressure in a brake wheel cylinder to carry out active pressure increase control, so that the vehicle is decelerated to a target speed;

the above processes are continuously carried out in the whole triggering process of the AEB system, and the information such as the real-time distance between the vehicle and the dangerous target, the early warning threshold values of all levels, the motion states of the vehicle and the dangerous target and the like is continuously updated and adjusted until the vehicle is decelerated to the target speed or stops and keeps a preset safe distance between the vehicle and the dangerous target.

2. The hierarchical automatic emergency brake control system for an automatically adjusting electric vehicle of claim 1, wherein: in the step (1), the method for obtaining the minimum road adhesion coefficient mu by the road adhesion coefficient estimation model based on filtering is as follows:

the Dugoff tire model is first built to obtain normalized tire forces, and represents the longitudinal and lateral forces on the tire as:

in the formula F_xIs a tire longitudinal force, F_yFor lateral forces of the tire, F_zIs the tire normal force, λ is the longitudinal slip ratio, C_yFor the cornering stiffness of the tyre, C_xα is tire slip angle, mu is road surface adhesion coefficient, epsilon is tire influence coefficient, is a parameter related to the structure and material of the tire and used for correcting the influence of vehicle slip speed on tire force, L is boundary value and used for describing the nonlinear characteristics of tire force caused by wheel slip, v is_xThe longitudinal speed of the whole vehicle; to facilitate the design of the road adhesion coefficient estimation algorithm, the Dugoff tire model is simplified to the following normalization form:

in the formula

And

the normalized tire forces in the longitudinal direction and the lateral direction are respectively independent of the adhesion coefficient mu, so that a coefficient matrix of a system state space expression is conveniently determined in the road adhesion coefficient estimation algorithm based on filtering;

after the normalized Dugoff tire model is established, the normalized tire force in the model needs to be calculated, namely the vertical load F of each wheel is calculated respectively_Zfl、F_Zfr、F_Zrl、F_ZrrLongitudinal slip ratio λ_fl、λ_fr、λ_rl、λ_rrTire slip angle α_fl，α_fr，α_rl，α_rrThe corner marks fl, fr, rl, rr represent the left front wheel, the right front wheel, the left rear wheel, and the right rear wheel of the vehicle, respectively;

the vertical load calculation formula of each wheel is as follows:

the formula for calculating the slip angle of each wheel is as follows:

in the formula: r is the vehicle yaw rate;

calculation formula of each wheel slip ratio:

velocity v of each wheel grounding point in the above formula_fl、v_fr、v_rl、v_rrThe calculation formula is as follows:

the meaning of each parameter in the above formulas is β is the vehicle mass center slip angle,

v_cogis the speed of the center of mass of the vehicle,

m is the total vehicle mass, a is the horizontal distance between the center of the front wheel and the vehicle mass center, b is the horizontal distance between the center of the rear wheel and the vehicle mass center, l is the vehicle wheel base, h_gIs the height of the center of mass of the vehicle, a_xFor longitudinal acceleration of the vehicle, a_yFor lateral acceleration of the vehicle, T_fIs a track of two front wheels, T_rIs a two rear wheel track v_xLongitudinal speed, v, of the entire vehicle_yLateral speed of the entire vehicle, R_ωAs wheel rolling speed, ω_fl、ω_fr、ω_rl、ω_rrRolling angular velocities of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel are respectively;

the longitudinal normalized tire force for the four wheels can thus be expressed as:

the lateral normalized tire force for the four wheels can be expressed as:

f(L_fl)、f(L_fr)、f(L_rl)、f(L_rr) Non-linear characteristic descriptors of tire acting force brought by slippage of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel; parameters required in the calculation process of the normalized tire force corrected by combining the Dugoff tire model comprise vehicle structure parameters and kinematic parameters, the structure parameters can be directly measured, and the kinematic parameters can be measured by a related vehicle-mounted sensor;

in order to write a system state equation and an observation equation in a filtering-based road adhesion coefficient estimation algorithm in a row, a three-degree-of-freedom four-wheel vehicle model of a vehicle is required to be established to describe the longitudinal, lateral and yaw motions of the vehicle;

the motion equations of the vehicle in the longitudinal direction, the lateral direction and the yaw direction are as follows:

in the formula, a_xFor longitudinal acceleration of the vehicle, a_yIn order to provide for a lateral acceleration of the vehicle,

is the yaw angular acceleration of the vehicle, delta is the two front wheel corners, m is the total vehicle mass, mu_fl、μ_fr、μ_rl、μ_rrRespectively, the road surface adhesion coefficients of a left front wheel, a right front wheel, a left rear wheel and a right rear wheel, a is the horizontal distance between the center of the front wheel and the mass center of the vehicle, b is the horizontal distance between the center of the rear wheel and the mass center of the vehicle, and T is the horizontal distance between the center of the rear wheel and the mass center of the vehicle_fIs a track of two front wheels, T_rTwo rear wheel track, I_zYaw moment of inertia of the vehicle;

the method comprises the following steps that related sensors on a vehicle measure vehicle kinematic parameters required by an algorithm and transmit the vehicle kinematic parameters to a Dugoff tire model, the Dugoff tire model calculates longitudinal and lateral normalized tire forces and transmits the longitudinal and lateral normalized tire forces to a vehicle three-degree-of-freedom four-wheel model, so that three dynamic equations in the longitudinal direction, the lateral direction and the yaw direction are obtained, and a system state equation and an observation equation of a road adhesion coefficient filter estimator can be obtained based on the three dynamic equations to carry out subsequent adhesion coefficient estimation;

the system selected by the road adhesion coefficient estimation algorithm based on filtering is a nonlinear system, and the state equation and the observation equation of the system are respectively as follows:

the state equation is as follows:

the observation equation:

y(t)＝h(x(t),u(t),v(t))

the random variables w (t), v (t) are respectively process noise and measurement noise, Gaussian white noise which are independent from each other and have a zero mean value is taken in the filtering-based road adhesion coefficient estimation algorithm, and the probability distribution is as follows:

p(w)～N(0,Q)p(v)～N(0,R)

and setting their covariance matrices as Q and R, respectively, namely:

Q＝cov[w(t),w(τ)]

R＝cov[v(t),v(τ)]

the specific implementation process of the road adhesion coefficient estimation algorithm based on filtering is as follows:

(1) determining a system state equation and an observation equation:

determining according to the motion equations of the longitudinal direction, the lateral direction and the yaw direction of the three-degree-of-freedom four-wheel vehicle model:

the state variables are as follows: x (t) ═ μ_flμ_frμ_rlμ_rr]^T，

Observation variables: y (t) ═ a_xa_yr]^T，

And (3) control input: u (t) ═ δ,

writing a state equation and an observation equation into the three-degree-of-freedom four-wheel vehicle model according to expressions and variables of motion equations in the longitudinal direction, the lateral direction and the yaw direction of the three-degree-of-freedom four-wheel vehicle model:

the state equation is as follows:

the observation equation:

wherein:

(2) the estimator assigns an initial value:

measuring an initial value in a recursion process, wherein a noise covariance matrix R is a 3-by-3 unit matrix, and a process noise covariance matrix Q is a 4-by-4 unit matrix;

(3) estimating the road adhesion coefficient by using a filtering algorithm:

①, the filter is initialized, the iteration step number k is 0, and the initial mean value of the state variable x

And the covariance P (0) are:

② calculating by using unscented transformation to obtain 2n +1 points, making the mean and covariance of the 2n +1 points equal to those of the original state distribution, and referring the 2n +1 points as Sigma points, n as state dimension, according to the above state equation, taking n as 4, then obtaining 9 Sigma point sets, where the state dimension of each Sigma point set is consistent with that of state variable x, that is, 4-dimensional column vector, then the matrix x formed by the 9 Sigma point sets is 4 x 9 matrix, and setting χ_i(0) For each column vector, i.e. each Sigma point set, χ in the matrix χ at iteration step k ═ 0₀(0) Column one of Sigma matrix χ, and so on, the initial Sigma matrix is:

where i is the number of each column vector in the matrix χ, λ is a scaling parameter, and λ is (α)²-1)·n＝4(α²-1), α for determining the Sigma point at the mean of the state variable

The nearby distribution range is a very small positive number 10^-4≤α≤1；

The above-mentioned calculation process of the initial Sigma point set when the iteration step number k is 0 is also applicable to the Sigma point set calculation of other iteration step numbers, except that the state variable mean value and the state variable covariance value used for the Sigma point set calculation at this time are values at the time k, that is, values at the time k

And P (k | k), the calculation formula is:

wherein i is the number of each column vector in the matrix χ;

③, proceeding to the next iteration, increasing the iteration step number k by 1, predicting the 9 Sigma point sets in the next step, substituting the Sigma point set obtained in the step ② into the state equation (a) of the system to obtain a transformed Sigma point set:

χ(k|k-1)＝f(χ(k-1),u(k-1))k＝1,2,…

④, calculating the next prediction mean value and the prediction variance of the system state variable according to the principle of unscented transformation, wherein the prediction mean value of the system state variable is obtained by the weighted summation of the prediction values of the Sigma point set:

in the formula W_i ^(m)The average weight of each Sigma point is specifically:

W₀ ^(m)＝λ/(4+λ)i＝0

W_i ^(m)＝1/[2(4+λ)]i＝1,2,…,8

wherein i is the number of each column vector in the matrix χ;

the predicted variance of the system state variables is obtained by weighted summation of the predicted covariance of the Sigma point set:

middle x of the above formula_i(k | k-1) is the ith column of the matrix χ (k | k-1), i ═ 0,1, …, 8; w_i ^(c)The covariance weight of each Sigma point is specifically:

W₀ ^(c)＝λ/(n+λ)+(1-α²+β)i＝0

W_i ^(c)＝1/[2(n+λ)]i＝1,2,…,8

wherein i is the number of each column vector in the matrix χ;

β, wherein the state variables x in the filter-based road adhesion coefficient estimation algorithm obey Gaussian distribution, and β is 2;

⑤ generating a new Sigma point set by using the unscented transformation again according to the next predicted mean and the next predicted variance of the system state variables obtained in step ④, the calculation process is the same as that of the initial Sigma point set in step ②:

wherein i is the number of each column vector in the matrix χ;

⑥ substituting the new Sigma point set obtained in step ⑤ into observation equation (b) to obtain the next predicted value of the observed quantity of the Sigma point set:

u (k-1) in the formula is system input at the moment when the iteration step number is k-1, namely the front wheel rotation angle delta at the moment can be measured by the existing vehicle-mounted rotation angle sensor;

by observation ofAs can be seen from equation (b), the output matrix of the system is a 3-by-4 matrix, and the Sigma matrix χ (k | k-1) is a 4-by-9 matrix, so that the next predicted value of the observed quantity of the Sigma point set is calculated through observation equation (b)

A matrix of 3 by 9;

⑦ predicted value from the observed value of Sigma Point set obtained in step ⑥

Obtaining a predicted mean value of the system observation variable through weighted summation:

in the formula

Matrix of next prediction values of observed quantities of Sigma point set

I ═ 0,1, …, 8;

⑧ predicted value from the observed value of Sigma Point set obtained in step ⑥

And predicted mean of system observed variables

Calculating an updated observation covariance matrix and a state variable and output variable cross-correlation matrix by weighted summation:

the observed covariance matrix is:

the state variable and output variable cross-correlation matrix is:

⑨ calculating the updated filtered feedback gain matrix of the system:

calculating the updated state variable mean matrix of the system:

matrix array

The (1,1) element, (2,1) element, (3,1) element, and (4,1) element are the filter estimation values of the left front wheel adhesion coefficient, the right front wheel adhesion coefficient, the left rear wheel adhesion coefficient, and the right rear wheel adhesion coefficient when the iteration step number is k, respectively;

wherein y (k) is the actual observed variable value at the moment when the iteration step number is k, i.e. the longitudinal acceleration a at the moment when the iteration step number is k_xLateral acceleration a_yAnd yaw angular acceleration

The three vehicle kinematic parameters can be measured in real time by a relevant acceleration sensor assembled on the vehicle and then transmitted to the road adhesion coefficient estimation algorithm based on filtering;

calculating the updated state variable covariance value of the system:

P(k|k)＝P(k|k-1)-K(k)P_yyK^T(k)

mean value of state variable after system update

And the state variable covariance value P (k | k) returns to step ② to generate a next group of Sigma point sets, the calculation of the next iteration step is started, the process is repeated continuously until all the iteration steps are completed, the road adhesion coefficients mu of the four wheels are finally obtained, the adhesion coefficient values of the four wheels are compared to obtain the minimum value, and then the minimum value mu is transmitted to the safe distance calculation model.

3. The hierarchical automatic emergency brake control system for an automatically adjusting electric vehicle of claim 1, wherein: in the step (4), the method for estimating the gradient i of the road surface where the vehicle is located in the road surface information estimation model based on the road surface gradient estimation algorithm of the closed-loop full-dimensional state observer is as follows:

the related sensors of the vehicle transmit vehicle parameters required by road gradient estimation to a road information estimation model, and the vehicle running equation is F_t＝F_f+F_w+F_i+F_j；

Vehicle driving force F_tComprises the following steps:

where r is the rolling radius of the wheel, T_eAs vehicle engine torque, i_gFor the transmission ratio of the vehicle transmission, i₀As a vehicle final drive, η_tThe overall efficiency of the vehicle driveline;

air resistance F of vehicle_wComprises the following steps:

wherein C is_DIs the coefficient of air resistance, A is the frontal area, ρ is the air density, v_xIs the vehicle longitudinal speed;

vehicle acceleration resistance F_jComprises the following steps:

F_j＝δMa_x

wherein delta is a conversion coefficient of rotating mass of vehicle(ii) a M is the total vehicle mass, a_xIs the vehicle longitudinal acceleration;

rolling resistance F of vehicle_fComprises the following steps:

F_f＝Mgf cosα≈Mgf

wherein f is the vehicle rolling resistance coefficient;

gradient resistance F of vehicle_iComprises the following steps:

F_i＝Mg sinα≈Mgi

the longitudinal dynamic equation of the vehicle on the slope can be written as:

driving force F of vehicle_tAir resistance F_wAnd rolling resistance F_fViewed as a resultant force F_inputAs system input of the vehicle longitudinal dynamics model, the longitudinal dynamics equation is simplified as follows:

δMa＝F_input-Mgi

the above formula is a system state equation, and a state space expression of the system can be written based on the system state equation; at vehicle longitudinal speed v_xAnd gradient i as system state variable, with resultant force F_inputFor system input variables, the state space expression based on the automobile longitudinal dynamics equation is as follows:

z＝Cx

wherein:

the observability matrix of the system is:

observability matrix Q_BFull rank, accounting for the systemTherefore, the road surface gradient observer is designed as follows:

wherein

Is an observation vector of the observer,

h is the feedback gain matrix of the observer; e is an error vector, i.e.

Will be provided with

And

subtraction can give:

is provided with

J, the above formula is changed to

Is converted into a first-order linear differential equation about j, and the initial time is set as t₀Then the differential equation is solved as:

to guarantee observer stability, it should be guaranteed that the observer error vector e is equal to 0 when the time goes to infinity, i.e.:

as long as the characteristic root λ of (a-HC) has a negative real part, the error vector will decay exponentially to 0, and the decay rate is determined by the characteristic root of (a-HC); setting feedback gain matrix of observer

The characteristic equation of (A-HC) is:

the expected pole of the full-dimensional observer, i.e. the (A-HC) characteristic root with a negative real part, is preset to be lambda₁And λ₂Expectation pole λ₁And λ₂The corresponding expected characteristic equation is:

(λ-λ₁)(λ-λ₂)＝λ²-(λ₁+λ₂)λ+λ₁λ₂(5)

let the elements of the feedback gain matrix H of the available observer with the same term coefficients for equations (4) and (5) be:

h₁＝-λ₁-λ₂

the characteristic value lambda can be ensured by calculating and obtaining the feedback gain matrix H according to the method₁And λ₂The observer can be stabilized by having a negative real part; full-dimensional state observer observation variable

Second row element of (1)

I.e. the estimated value i of the road gradient on which the vehicle is located at the moment.

Images