CN106218633A - Four motorized wheels electric automobile stability control method based on Q study - Google Patents

Abstract

The invention discloses a kind of four motorized wheels electric automobile stability control method based on Q study, comprise the following steps: select corresponding optimal control parameter according to actual external condition, and utilize these parameters to be calculated preferable control moment；Calculate the sliding formwork in the yaw moment control device under different external condition and control parameter K, and sliding formwork control parameter K in the yaw moment control device of different external conditions and correspondence thereof is stored in stabilitrak；Calculated preferable control moment is reasonably assigned to four wheels.By the way of Q learns, find the control parameter needed for line computation and be stored in yaw moment control device, the yaw moment control device of four motorized wheels electric automobile operationally can directly invoke control parameter in the way of tabling look-up, this substantially reduces the calculating time, improves the real-time of four motorized wheels electric automobile stabilitrak, robustness and practicality.

Description

Four motorized wheels electric automobile stability control method based on Q-study

Technical field

The invention belongs to automotive safety auxiliary drive and field of intelligent control, relate to four motorized wheels electric automobile control The method for designing of system processed, is related specifically to four motorized wheels electric automobile stability control method.

Background technology

Along with the development of society, energy problem is more and more severeer, and traditional combustion engine automobile can not meet people to low The demand that carbocyclic ring keeping normal life activities is lived.Electric automobile energy saving is pollution-free, is environmentally friendly vehicle truly.The electronic vapour of four motorized wheels Car, as an important branch of electric automobile, is increasingly paid close attention to by people.Four motorized wheels electric automobile achieves The electronization of automobile chassis, being arranged on four motors near wheel or in wheel hub can independently control, and intelligent algorithm is transported With, it is easier to realize the intellectuality of automobile.

Automobile stability control system develops on the basis of anti-locking system for car, and its Main Function is Under the limiting condition of big lateral acceleration and big side slip angle, come by control action longitudinal force on four wheels Generation yaw moment, to prevent unmanageable sideslip phenomenon, makes automobile travel according to the wish of driver.At present, vehicle steadily Property control system be mainly used on orthodox car, automobile stability control system by control four wheels braking moment produce Raw yaw moment, it is ensured that the stability of automobile.For four motorized wheels electric automobile, automobile stability control system is by control The output torque of four motors of system produces yaw moment, it is ensured that the stability of automobile.Four-wheeled independent driving automobile stability control The subject matter that system processed exists is: first, the currently design master to four motorized wheels electric automobile stabilitrak Intelligent control algorithm to be used, after using intelligent control algorithm, owing to the intelligent control algorithm calculating cycle is long, four-wheel is independent The real-time driving electric automobile stabilitrak is deteriorated.Second, currently four motorized wheels electric automobile is stably controlled The control algolithm parameter being designed with of system processed mostly is single, when the external condition of automobile changes, and vehicle steadily Property control system cannot be according to external condition auto-changing control algolithm parameter, and robustness and the practicality of control system are bad.

Summary of the invention

For solving the problems referred to above that prior art exists, present invention one to be designed can improve electric automobile stability control The real-time of system processed, can improve again the robustness of electric automobile stabilitrak and based on Q-study the four of practicality Wheel is independent drives electric automobile stability control method.

To achieve these goals, technical scheme is as follows: a kind of four motorized wheels electricity based on Q-study Electrical automobile stability control method, comprises the following steps:

A, in line computation

Select corresponding optimal control parameter according to actual external condition, and utilize these parameters to be calculated preferably Control moment, specifically comprises the following steps that

A1, set up kinetic model

Set up the kinetic model of automobile i.e. according to the stressing conditions of automobile, set up the kinetics equation of automobile；Neglect Resistance to rolling, air drag, grade resistance and the impact of automobile catenary motion, four motorized wheels electric automobile longitudinally, Horizontal and yaw direction kinetics equation is expressed as follows:

$m(\stackrel{\·}{V_x}-V_y\mathrm{\γ})=(F_{x1}+F_{x2}){\mathrm{cos\δ}}_f+F_{x3}+F_{x4}-(F_{y1}+F_{y2}){\mathrm{sin\δ}}_f---\left(1\right)$

$m({\stackrel{\·}{V}}_y+V_x\mathrm{\γ})=(F_{x1}+F_{x2}){\mathrm{sin\δ}}_f+(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f+F_{y3}+F_{y4}---\left(2\right)$

$I_z\stackrel{\·}{\mathrm{\γ}}=l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2}){\mathrm{sin\δ}}_f+M_x---\left(3\right)$

Wherein, m is the quality of automobile, l_fFor the distance of front axle to barycenter, l_rFor the distance of rear axle to barycenter, l_wFor automobile Distance between left and right wheels；I_zFor automobile rotary inertia in barycenter yaw direction, r is the effectively rolling half of automobile tire Footpath, F_x1、F_x2、F_x3、F_x4It is respectively the longitudinal force suffered by the near front wheel, off-front wheel, left rear wheel, off hind wheel tire, F_y1、F_y2、F_y3、 F_y4It is respectively the side force suffered by the near front wheel, off-front wheel, left rear wheel, off hind wheel tire, δ_fFor vehicle front corner, γ is automobile Yaw velocity,For the yaw angle acceleration of automobile, V_xFor longitudinal speed of automobile,For the longitudinal acceleration of automobile, V_y For the lateral speed of automobile,Lateral acceleration for automobile.M_xFor direct yaw moment, by the output torque phase of four motors Interaction produces, and its expression formula is:

$M_x=\frac{l_w}{2}(F_{x2}-F_{x1}){\mathrm{cos\δ}}_f+l_f(F_{x1}+F_{x2}){\mathrm{sin\δ}}_f+\frac{l_w}{2}(F_{x4}-F_{x3})---\left(4\right)$

A2, design yaw moment control device

During in view of external disturbance, the weaving equation below of automobile represents:

$I_z\stackrel{\·}{\mathrm{\γ}}=l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2}){\mathrm{sin\δ}}_f+M_x+M_d---\left(5\right)$

Wherein, M_dFor the disturbance torque produced by the lateral wind suffered by automobile and Uneven road；M_dThere is border, represent For:

M_d≤D

Wherein, D is the coboundary of disturbance torque.

The mode using sliding formwork to control designs yaw moment control device, and the sliding-mode surface chosen is defined as:

S=γ-γ_d

Wherein, γ is the yaw velocity of automobile, γ_dPreferable yaw velocity for automobile.

The computing formula obtaining direct yaw moment according to sliding formwork control principle is as follows:

$M_x=I_z{\stackrel{\·}{\mathrm{\γ}}}_d-\{l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2})\}+I_{z}K\mathrm{sgn}\left(s\right)---\left(6\right)$

Wherein, K is that the sliding formwork in yaw moment control device controls parameter, is formulated as:

K ＞ D/I_z

The mode using calculated off line controls parameter K to sliding formwork and carries out value.

B, calculated off line

The sliding formwork calculated in the yaw moment control device under different external condition controls parameter K, and by difference external condition And the sliding formwork in the yaw moment control device of correspondence controls parameter K and stores in stabilitrak, specifically comprise the following steps that

B1, generation primary, initialize population, population quantity be set to n, and the dimension of each particle is set to 1；

B2, by the position numerical value assignment of primary to yaw moment control device sliding formwork control parameter K；

B3, sliding formwork controls parameter K being delivered to yaw moment control device, yaw moment control device is according to longitudinal speed and ground Face attachment coefficient calculates the size of direct yaw moment；

B4, calculated direct yaw moment is acted on four motorized wheels electric vehicle simulation model, calculate reason Think the difference e (t) of yaw velocity and actual yaw velocity；

The least fitness representing particle is the best with the difference of actual yaw velocity for B5, preferable yaw velocity, chooses The integration of squared difference is as evaluation function:

$f={\∫}_0^{+\mathrm{\∞}}{e}^2\left(t\right)dt$

Utilize this evaluation function that the fitness value of each particle is calculated, find individual extreme value and colony's extreme value；

B6, end condition be: maximum iteration time is taken as 250-350 or iteration precision threshold value takes 10^-4-10^-3.When meeting End condition stops during any one iteration, sliding formwork control parameter K in output yaw moment control device, and in a tabular form The longitudinal speed of storage, ground attaching coefficient and sliding formwork control parameter K to yaw moment control device；Go to step B11；If being unsatisfactory for end Only condition, then enter into step B7；

B7, by Q-study adjust inertial factor ω.Choose the position of current particle and distance d of its individual extreme value₁With The position of current particle and distance d of its colony's extreme value₂Two amounts, as the state of Q-study, choose inertial factor ω as Q- The action of study, the span of inertial factor ω is [ω_min,ω_max], it is averaged discrete for n part, obtains discrete inertia The span of factor ω is [ω₁,ω₂,…,ω_n], by ω₁、ω₂、…、ω_nAs n action of Q-study, below employing Step carries out Q-study, finds and adjusts inertial factor ω:

B71, for each particle, initialize Q-and learn formEach element is initialized as 0；

B72, observe current state s, repeat to do always:

B721, one action a of selection also perform it, and a is ω₁、ω₂、…、ω_nIn any one；

B722, receive and return r immediately；

B723, observation new state s '；

B724, to Q-learn formUpdate according to formula (7)

$\hat{Q}(s,a)\←r+\mathrm{\γ}\underset{a^{\′}}{max}\hat{Q}(s^{\′},a^{\′})---\left(7\right)$

Wherein,For the updated value after state s execution action a, r is returning immediately after state s execution action a Report value, γ is commutation factor,Represent: agent performs the corresponding maximum Q-value of next action a ' when state s '. Agent is the agency in Q-study.

B725, renewal current state: s ← s '

B8, through successive ignition, whenWhen converging to changeless Q, iteration stopping, optimal strategy produces, chooses Optimum action inertia factor ω corresponding under current state is as adjusting inertial factor ω；

B9, each particle update speed and the position of oneself according to formula (8), (9)；

$V_{id}^{k+1}={\mathrm{\ωV}}_{id}^k+c_1{r}_1(P_{id}^k-X_{id}^k)+c_2{r}_2(P_{gd}^k-X_{id}^k)---\left(8\right)$

$X_{id}^{k+1}=X_{id}^k+V_{id}^{k+1}---\left(9\right)$

Wherein, V_idFor the speed of particle, ω is inertial factor, and k is current iteration number of times, c₁And c₂It is the constant of non-negative, c It is referred to as acceleration factor, r₁And r₂It is distributed across the random number that [0,1] is interval, P_idFor individual extreme value, P_gdColony pole for population Value, X_idPosition for particle.

B10, the numerical value assignment of position step B9 updated control parameter K to the sliding formwork in yaw moment control device, return Return to step B3 and enter next cycle of operation.

B11, finally storage will control parameter K to longitudinal speed of yaw moment control device, ground attaching coefficient and sliding formwork Represent by following table:

Table 1: the sliding formwork in yaw moment control device controls parameter K

Wherein K_ijFor being V when speed_i, ground attaching coefficient be μ_iTime yaw moment control device in sliding formwork control parameter The numerical value of K, i=1,2 ..., n, j=1,2 ..., n.

C, Torque distribution

According to following Torque distribution rule calculated preferable control moment is reasonably assigned to four wheels:

Yaw moment control device, according to current speed and ground attaching coefficient, inquires the sliding formwork control of correspondence from table 1 Parameter K processed, and sliding formwork is controlled parameter K be transported to formula (6) and calculate direct yaw moment M_x, and carry out moment according to the following steps Distribution:

If the torque command of driver's input is T_driver, yaw moment control device calculates the direct yaw moment of gained and is M_x, the moment difference of coaxial left-right wheel is Δ T=T_right-T_left, the radius of clean-up of four wheels is r.Torque difference between wheel The yaw moment that the yaw moment that value produces produces equal to yaw moment control device, represents by equation below:

$\mathrm{\Δ}T\frac{l_w}{r}=M_x---\left(11\right)$

Select T_driverMoment on the basis of/4, is adjusted the moment of four wheels on the basis of benchmark moment so that Four wheels can meet driver requested driving moment, can produce again the moment difference of needs, adjusts formula as follows:

$T_1=T_3=\frac{T_{driver}}{4}-\frac{{\mathrm{rM}}_{desire}}{2l_w}---\left(12\right)$

$T_2=T_4=\frac{T_{driver}}{4}+\frac{{\mathrm{rM}}_{desire}}{2l_w}---\left(13\right)$

Wherein, T₁、T₂、T₃、T₄Respectively it is assigned to the near front wheel correspondence motor, off-front wheel correspondence motor, left rear wheel corresponding The moment of the motor that motor, off hind wheel are corresponding.

Compared with prior art, the method have the advantages that

1, the present invention proposes a kind of four motorized wheels electric automobile stability control method based on machine learning, bag Including at line computation, calculated off line and Torque distribution, calculated off line finds the control needed for line computation by the way of Q-learns Parameter processed is also stored in yaw moment control device, and the yaw moment control device of four motorized wheels electric automobile is in work Time can directly invoke control parameter in the way of tabling look-up, this substantially reduces the calculating time, improve four motorized wheels electricity The real-time of electrical automobile stabilitrak.

The sliding formwork that the present invention uses particle swarm optimization algorithm to find in optimum yaw moment control device controls parameter K.Research Show, although particle cluster algorithm has many good qualities, but be easily trapped into local optimum.In order to avoid particle is absorbed in local optimum, Improving the quality of particle cluster algorithm, the method that the present invention uses Q-to learn dynamically adjusts the inertial factor in particle swarm optimization algorithm ω。

2, particle swarm optimization algorithm is computational intelligence field, the kind of groups intelligence in addition to ant group algorithm, fish-swarm algorithm Can optimized algorithm.Traditional parameter testing determines the size of parameter mainly by the mode of expertise or test trial and error, this Planting debud mode to waste time and energy, especially when relating to multiple parameter and debugging simultaneously, traditional parameter testing mode is difficult to look for To optimal parameter combination.The calculated off line of the present invention is learnt by Q-and particle cluster algorithm is found in yaw moment control device Sliding formwork controls parameter K, and this is more time-consuming than the mode relying on trial and error and dependence expertise, and the optimal controller found Parametric reliability is strong.

3, the present invention proposes to find optimal control parameter under different external conditions, then by difference external condition and right The optimal control parameter answered stores in stabilitrak, when the external condition of automobile changes, and Vehicle Stability Control System selects corresponding optimal control parameter according to different external condition, improves robustness and the practicality of control system.

4, this based on machine learning the four motorized wheels electric automobile stability control method pair that the present invention proposes The stabilitrak design of orthodox car has certain reference significance.

Accompanying drawing explanation

The present invention has 2, accompanying drawing, wherein:

Fig. 1 is the overview flow chart of the present invention；

Fig. 2 is the calculated off line flow chart of the present invention.

Detailed description of the invention

Below in conjunction with the accompanying drawings the present invention is further described through.A kind of four motorized wheels based on Q-study is electronic Vehicle Stability Control method flow diagram is as it is shown in figure 1, wherein step B2 flow process is as shown in Figure 2.

In prior art, the design to four motorized wheels electric automobile stabilitrak mainly uses Based Intelligent Control Algorithm, uses after intelligent control algorithm, and it is long that intelligent control algorithm calculates the cycle, sometimes iteration one step need a few minutes time Between, this does not clearly meet the requirement to real-time of the four motorized wheels electric automobile stabilitrak.The present invention is to adopt By the mode of calculated off line, data are stored in Yaw stability controller in the way of form.When automobilism, online The data of storage only need to be directly invoked in the way of tabling look-up, it may be said that be Millisecond during calculating.

The present invention is not limited to the present embodiment, any equivalent concepts in the technical scope of present disclosure or change Become, be all classified as protection scope of the present invention.

Claims (1)

1. a four motorized wheels electric automobile stability control method based on Q-study, it is characterised in that: include following Step:

A, in line computation

Select corresponding optimal control parameter according to actual external condition, and utilize these parameters to be calculated preferably control Moment, specifically comprises the following steps that

A1, set up kinetic model

Set up the kinetic model of automobile i.e. according to the stressing conditions of automobile, set up the kinetics equation of automobile；Neglect rolling Resistance, air drag, grade resistance and the impact of automobile catenary motion, four motorized wheels electric automobile is longitudinally, laterally And the kinetics equation of yaw direction is expressed as follows:

$m({\stackrel{\·}{V}}_x-V_y\mathrm{\γ})=(F_{x1}+F_{x2}){\mathrm{cos\δ}}_f+F_{x3}+F_{x4}-(F_{y1}+F_{y2}){\mathrm{sin\δ}}_f---\left(1\right)$

$m({\stackrel{\·}{V}}_y+V_x\mathrm{\γ})=(F_{x1}+F_{x2}){\mathrm{sin\δ}}_f+(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f+F_{y3}+F_{y4}---\left(2\right)$

$I_z\stackrel{\·}{\mathrm{\γ}}=l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2}){\mathrm{sin\δ}}_f+M_x---\left(3\right)$

Wherein, m is the quality of automobile, l_fFor the distance of front axle to barycenter, l_rFor the distance of rear axle to barycenter, l_wFor vehicle right and left Distance between wheel；I_zFor automobile rotary inertia in barycenter yaw direction, r is the effective rolling radius of automobile tire, F_x1、F_x2、F_x3、F_x4It is respectively the longitudinal force suffered by the near front wheel, off-front wheel, left rear wheel, off hind wheel tire, F_y1、F_y2、F_y3、F_y4Point The not side force suffered by the near front wheel, off-front wheel, left rear wheel, off hind wheel tire, δ_fFor vehicle front corner, γ is the horizontal stroke of automobile Pivot angle speed,For the yaw angle acceleration of automobile, V_xFor longitudinal speed of automobile,For the longitudinal acceleration of automobile, V_yFor vapour The lateral speed of car,Lateral acceleration for automobile；M_xFor direct yaw moment, by the output torque phase interaction of four motors With generation, its expression formula is:

$M_x=\frac{l_w}{2}(F_{x2}-F_{x1}){\mathrm{cos\δ}}_f+l_f(F_{x1}+F_{x2}){\mathrm{sin\δ}}_f+\frac{l_w}{2}(F_{x4}-F_{x3})---\left(4\right)$

A2, design yaw moment control device

During in view of external disturbance, the weaving equation below of automobile represents:

$I_z\stackrel{\·}{\mathrm{\γ}}=l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2}){\mathrm{sin\δ}}_f+M_x+M_d---\left(5\right)$

Wherein, M_dFor the disturbance torque produced by the lateral wind suffered by automobile and Uneven road；M_dThere is border, be expressed as:

M_d≤D

Wherein, D is the coboundary of disturbance torque；

The mode using sliding formwork to control designs yaw moment control device, and the sliding-mode surface chosen is defined as:

S=γ-γ_d

Wherein, γ is the yaw velocity of automobile, γ_dPreferable yaw velocity for automobile；

The computing formula obtaining direct yaw moment according to sliding formwork control principle is as follows:

$M_x=I_z{\stackrel{\·}{\mathrm{\γ}}}_d-\{l_f(F_{y1}+F_{y2}){\mathrm{cos\δ}}_f-l_r(F_{y3}+F_{y4})+\frac{l_w}{2}(F_{y1}-F_{y2})\}+I_{z}K\mathrm{sgn}\left(s\right)---\left(6\right)$

Wherein, K is that the sliding formwork in yaw moment control device controls parameter, is formulated as:

K ＞ D/I_z

The mode using calculated off line controls parameter K to sliding formwork and carries out value；

B, calculated off line

Calculate the sliding formwork in the yaw moment control device under different external condition and control parameter K, and by difference external condition and The corresponding sliding formwork in yaw moment control device controls parameter K and stores in stabilitrak, specifically comprises the following steps that

B1, generation primary, initialize population, population quantity be set to n, and the dimension of each particle is set to 1；

B2, by the position numerical value assignment of primary to yaw moment control device sliding formwork control parameter K；

B3, sliding formwork controls parameter K being delivered to yaw moment control device, yaw moment control device is attached according to longitudinal speed and ground The size of coefficient calculations direct yaw moment；

B4, calculated direct yaw moment is acted on four motorized wheels electric vehicle simulation model, calculate preferable horizontal Pivot angle speed and the difference e (t) of actual yaw velocity；

The least fitness representing particle is the best with the difference of actual yaw velocity for B5, preferable yaw velocity, chooses difference Integrated square is as evaluation function:

$f={\∫}_0^{+\mathrm{\∞}}{e}^2\left(t\right)dt$

Utilize this evaluation function that the fitness value of each particle is calculated, find individual extreme value and colony's extreme value；

B6, end condition be: maximum iteration time is taken as 250-350 or iteration precision threshold value takes 10^-4-10^-3；When satisfied termination bar Stopping iteration in part during any one, the sliding formwork in output yaw moment control device controls parameter K, and storage is vertical in a tabular form Parameter K is controlled to yaw moment control device to speed, ground attaching coefficient and sliding formwork；Go to step B11；If being unsatisfactory for terminating bar Part, then enter into step B7；

B7, by Q-study adjust inertial factor ω；Choose the position of current particle and distance d of its individual extreme value₁With current grain The position of son and distance d of its colony's extreme value₂Two amounts, as the state of Q-study, choose what inertial factor ω learnt as Q- Action, the span of inertial factor ω is [ω_min,ω_max], it is averaged discrete for n part, obtains discrete inertial factor ω Span be [ω₁,ω₂,…,ω_n], by ω₁、ω₂、…、ω_nAs n action of Q-study, following steps are used to enter Row Q-learns, and finds and adjusts inertial factor ω:

B71, for each particle, initialize Q-and learn formEach element is initialized as 0；

B72, observe current state s, repeat to do always:

B721, one action a of selection also perform it, and a is ω₁、ω₂、…、ω_nIn any one；

B722, receive and return r immediately；

B723, observation new state s '；

B724, to Q-learn formUpdate according to formula (7)

$\hat{Q}(s,a)\←r+\mathrm{\γ}\underset{a^{\′}}{max}\hat{Q}(s^{\′},a^{\′})---\left(7\right)$

Wherein,For the updated value after state s execution action a, r is the return value immediately after state s execution action a, γ is commutation factor,Represent: agent performs the corresponding maximum Q-value of next action a ' when state s '；agent Agency in learning for Q-；

B725, renewal current state: s ← s '

B8, through successive ignition, whenWhen converging to changeless Q, iteration stopping, optimal strategy produces, and chooses current Optimum action inertia factor ω corresponding under state is as adjusting inertial factor ω；

B9, each particle update speed and the position of oneself according to formula (8), (9)；

$V_{id}^{k+1}={\mathrm{\ωV}}_{id}^k+c_1{r}_1(P_{id}^k-X_{id}^k)+c_2{r}_2(P_{gd}^k-X_{id}^k)---\left(8\right)$

$X_{id}^{k+1}=X_{id}^k+V_{id}^{k+1}---\left(9\right)$

Wherein, V_idFor the speed of particle, ω is inertial factor, and k is current iteration number of times, c₁And c₂Being the constant of non-negative, c is referred to as Acceleration factor, r₁And r₂It is distributed across the random number that [0,1] is interval, P_idFor individual extreme value, P_gdFor colony's extreme value of population, X_id Position for particle；

B10, the numerical value assignment of position step B9 updated control parameter K to the sliding formwork in yaw moment control device, return to Step B3 enters next cycle of operation；

B11, will finally storage control under parameter K to longitudinal speed of yaw moment control device, ground attaching coefficient and sliding formwork Table represents:

Table 1: the sliding formwork in yaw moment control device controls parameter K

Wherein K_ijFor being V when speed_i, ground attaching coefficient be μ_iTime yaw moment control device in sliding formwork control parameter K number Value, i=1,2 ..., n, j=1,2 ..., n；

C, Torque distribution

According to following Torque distribution rule calculated preferable control moment is reasonably assigned to four wheels:

Yaw moment control device controls ginseng according to current speed and ground attaching coefficient, the sliding formwork inquiring correspondence from table 1 Number K, and sliding formwork is controlled parameter K be transported to formula (6) and calculate direct yaw moment M_x, and carry out moment according to the following steps and divide Join:

If the torque command of driver's input is T_driver, it is M that yaw moment control device calculates the direct yaw moment of gained_x, with The moment difference of axle left and right wheels is Δ T=T_right-T_left, the radius of clean-up of four wheels is r；Between wheel, moment difference produces Yaw moment equal to yaw moment control device produce yaw moment, represent by equation below:

$\mathrm{\Δ}T\frac{l_w}{r}=M_x---\left(11\right)$

Select T_driverMoment on the basis of/4, is adjusted the moment of four wheels on the basis of benchmark moment so that four Wheel can meet driver requested driving moment, can produce again the moment difference of needs, adjusts formula as follows:

$T_1=T_3=\frac{T_{driver}}{4}-\frac{{\mathrm{rM}}_{desire}}{2l_w}---\left(12\right)$

$T_2=T_4=\frac{T_{driver}}{4}+\frac{{\mathrm{rM}}_{desire}}{2l_w}---\left(13\right)$

Wherein, T₁、T₂、T₃、T₄Respectively be assigned to the near front wheel correspondence motor, off-front wheel correspondence motor, motor that left rear wheel is corresponding, The moment of the motor that off hind wheel is corresponding.