Specific embodiment
Below in conjunction with the accompanying drawings, the technical solution proposed to invention is further elaborated and illustrates.
The present invention provides a kind of vehicle stability controller design method based on variable weight model prediction algorithm, such as Fig. 1
Shown, this method includes the following steps:
1. acquisition in real time and processing vehicle's current condition information:
First, the car status information that acquisition onboard sensor is fed back in real time, acquires including rotation direction sensor
The steering wheel angle signal come, the yaw velocity and angle of heel speed at vehicle centroid that deflection efficiency sensor (gyroscope) acquires
Signal, the acceleration signal longitudinally and laterally that acceleration transducer measures are spent, pressure sensor is detected to obtain and be braked at wheel
The actual pressure signal of pipe.By these signals are filtered or estimation processing, so as to which the current steering wheel of real-time update vehicle turns
Angle, yaw velocity, angle of heel, roll velocity, side slip angle, lateral load transfer, longitudinal speed and wheel cylinder braking
The status informations such as pressure.
2. with handling obtained car status information in real time according to step 1, the holding vehicle yaw stability that makes separate decisions out and
The reference state value of roll stability:
Since yaw velocity can really react the weaving state of vehicle, and with can be measured directly excellent
Gesture, therefore the present invention really reacts the driving intention of driver by calculating the value with reference to yaw velocity, as vehicle
Keep the reference state of Yaw stability.In this step, the yaw velocity r of reference is calculated by formula (1)ref:
Wherein:L is car gage, vxFor longitudinal speed, δfFor front wheel angle, k represents stability factor, is characterization automobile
The important parameter of steady-state response, the present invention in value be 0.0025s2/m2
For the roll motion of vehicle, the present invention selects the lateral load rate of transform (Lateral-transfer-ratio) to make
For the rollover index of automobile, referred to as LTR.It is defined as the vertical load of tire both sides difference with vertical load and ratio
Exhausted angle value, shown in expression such as formula (2):
Wherein, FzlRepresent the sum of vertical load of vehicle the near front wheel and left rear wheel, FzrRepresent vehicle off-front wheel and off hind wheel
The sum of vertical load.
From the expression-form of the lateral load rate of transform, it can be seen that LTR is a variable between 0 and 1.Work as LTR
Numerical value be 0 when, the vertical load of vehicle left side tire is equal with the vertical load of right side tire, illustrates vehicle at this time just
In stable traveling;When the numerical value of LTR is 1, the vertical load of vehicle left side tire or right side tire is 0, and wheel has
Leave the danger on ground, it is easy to rollover event occur.Therefore it is to keep vehicle roll stability that the present invention, which selects LTR=0,
Reference state value.
3. based on variable weight model prediction algorithm, with front wheel angle and tire brake force, variable, design vehicle are horizontal in order to control
Pendulum and roll stability integrated manipulator:
(1) system prediction model
For the weaving, lateral movement and roll motion of vehicle, simplified Three Degree Of Freedom auto model conduct is established
Prediction model.Hypothesis below is made first:
I does not consider the influence of wheel steering system, front wheel angle δfInput as system;
II ignores the effect of suspension, it is believed that displacement of the automobile along z-axis, the angle of heel around x-axis are with the pitch angle around y-axis
Zero;
III does not consider the influence of the variation of tyre load and ground tangential force to tire cornering characteristics;
IV ignores aerodynamic effect, and driving force is little;
V automobile is constant along the longitudinal velocity of x-axis.
Using iso standard vehicle axis system, coordinate origin is located at the barycenter of vehicle the present invention, and vehicle is along headstock to moving ahead
The direction sailed is set as positive direction of the x-axis, and horizontal is the positive direction of y-axis to the left, and z-axis positive direction is determined by right-hand screw rule.Due toVehicle centroid lateral deviation angular rate of change is obtained according to up to bright BELL'S THEOREMAbout resulting side force ∑ Fy, sideway resultant force
Square ∑ MzAnd roll resultant moment ∑ MxExpression formula:
Yaw velocity change rateAbout resulting side force ∑ Fy, sideway resultant moment ∑ MzAnd roll resultant moment ∑ MxTable
Up to formula:
Roll velocity change rateAbout resulting side force ∑ Fy, sideway resultant moment ∑ MzAnd roll resultant moment ∑ MxTable
Up to formula:
Wherein, D=Ixs-ms 2e2/m-IxzIxzs/Iz
In order to describe the relationship between side force of tire and side drift angle, the present invention selects linear tire model, it is assumed that left front
Wheel and off-front wheel, left rear wheel and off hind wheel have identical steering characteristic, then vehicle front-wheel lateral force FyfWith front-wheel side drift angle αf、
Trailing wheel lateral force FyrWith trailing wheel side drift angle αrRelational expression such as following formula (6) shown in:
Therefore tire resulting side force ∑ F is obtainedy, sideway resultant moment ∑ MzAnd roll resultant moment ∑ MxExpression formula such as formula
(7) shown in-(9):
∑Fy=Kfαfcosδf+Krαr (7)
∑Mz=Lf·Kfαfcosδf-Lr·Krαr+ΔMz (8)
In above formula, m is complete vehicle quality, msFor vehicle spring carried mass, LfFor front-wheel axle center to the distance of automobile barycenter, LrFor
Rear axis to vehicle centroid distance, e be vehicle spring carried mass barycenter to roll center distance.G is acceleration of gravity.
vxRepresent longitudinal direction of car speed, vyThe lateral speed of vehicle is represented, r is yaw velocity, and β is side slip angle, and φ is body of a motor car
Angle of heel, δfIt is front wheel angle.MzIt is yaw moment of the vehicle around z-axis, Δ MzFor additional yaw moment, MxIt is vehicle around x-axis
Moment resulting from sidesway.KφFor the roll stiffness of vehicle, CφInclination for vehicle damps, KfFor front-wheel cornering stiffness, KrFor trailing wheel lateral deviation
Rigidity.IzIt is complete vehicle quality around the rotary inertia of z-axis, IxzIt is complete vehicle quality around x, the product of inertia of z-axis, IxsFor vehicle spring carried mass
Around the rotary inertia of vehicle axis system x-axis, IxzsIt is vehicle spring carried mass around vehicle axis system x, the product of inertia of z-axis.
Do not depend on what driver inputted since electronic control unit can apply front-wheel one by wire-controlled steering system
Additional rotation angle is superimposed maneuverability, stability and track retention property so as to improve vehicle with the go to action of driver.Cause
This, additional front wheel angle can be superimposed as a controlled variable of control system with go to action, and final synthesis obtains preceding rotation
Angle δfAs shown in following formula (10):
δf=δfd+Δδf (10)
Wherein, Δ δfBe control system output additional front wheel angle, δfdIt is the steering wheel angle δ process steerings of driver
Mechanism is applied to the corner on front-wheel, δ and δfdRelational expression such as following formula (11) shown in.
δfd=δ/Isw (11)
In formula:IswFor steering gear ratio.
There is nonlinear cornering behavior, when lateral force is in saturation state, course changing control however, as tire
The lateral dynamic characteristic of vehicle cannot be directly changed.And differential braking is made by the way that four tires are applied with different longitudinal directions
Power, so as to generate the additional yaw moment Δ M around Full Vehicle System z-axiszEven if still have when lateral force saturation
Certain validity.Therefore, using front wheel angle and tire brake force simultaneously as control variable.Set tire brake force positive direction
The direction advanced with vehicle is on the contrary, formula (12) gives additional yaw moment Δ MzWith the brake force Δ F of four tiresxiPass
System.
Wherein, d1、d2Respectively vehicle wheel base, Δ FyflFor the near front wheel brake force, Δ FyfrFor off-front wheel brake force,
ΔFyrlFor left rear wheel brake force, Δ FyrrFor off hind wheel brake force.
The basic control targe of Yaw stability is tracking of the yaw velocity for reference value, the root of roll stability
This control targe is to reduce the lateral load rate of transform of vehicle, therefore it is controlled to select yaw velocity and the lateral load rate of transform
Output quantity.Since the vertical load of both sides tire is not easy to measure, dynamic lateral load transfer rate LTR expression is obtained by shifting onto
Formula is as follows:
Wherein, d=(d1+d2)/2 are averaged wheelspan for front and back wheel, hRFor height of roll center.
From formula (13) as can be seen that LTR with β, r, φ andVariation and change.Therefore it is assumed herein that angle of heel
Change rate isAnd it is state variable i.e. x to select side slip angle, yaw velocity, angle of heel and inclination angular rate of change
=[β, r, φ, p], yaw velocity and the lateral load rate of transform are y=[r as controlled output;LTR], it will additional preceding rotation
Angle and tire brake force are as control input i.e. u=[Δ δf,ΔFxfl,ΔFxfr,ΔFxrl,ΔFxrr].Due to δfdIt is by driving
Member's output determines, it is impossible to it be predicted, therefore in the entire prediction time domain in each sampling period, it is believed that δfdIt is normal
Value.
Composite type (3)-(13), the continuous time state space equation expression formula that final arrangement obtains prediction model are as follows:
For the design of subsequent controllers, need, by continuous state-space model discretization, to select the sampling time as Ts
=0.01, after discretization, state-space model can be described as formula (15):
Here definition prediction time domain is p, and control time domain is m, and p>m.Since the variation of longitudinal speed is continuous and slow
Slow, therefore assume that speed is steady state value in prediction time domain.It is adapted with the rule of Nonlinear Model Predictive Control, it is assumed that vehicle
Current time is k, and dynamic can be based on vehicle's current condition to vehicle in [k+1, k+p] prediction time domain and prediction model obtains
It arrives.I.e. at the k+p moment, vehicle-state is x (k+p)=F (x (k), u (k), u (k+1) ..., u (k+m) ..., u (k+p-1)).
When being more than control time domain m in the sampling time, keep control input constant until predicting time domain, i.e. u (k+m-1)=u
(k+m)=u (k+m+1)=...=u (k+p-1).
Therefore the optimum control input at k moment is defined:
Correspondingly, the prediction output of k moment yaw velocities r can be calculated by above-mentioned state equation (17):
Equally, the lateral load rate of transform LTR prediction outputs at k moment are as follows:
It is corresponding with prediction output equation (17), it should predict real-time update in time domain each with reference to yaw velocity,
Reference input sequence definition is as follows:
(2) based on the feedback optimized of Fuzzy strategy
Consider the Yaw stability of vehicle and roll stability demand, the yaw velocity r of vehicle reality should use up
Yaw velocity r is referred in fast trackingref, while the reduction that lateral load rate of transform LTR should be as possible, therefore obtain integrated control
Shown in the object function of device processed such as following formula (20):
J=Γy||(Y1(k+1|k)-R(k+1))||2+Γr||(Y2(k+1|k))||2 (20)
Herein, Γy=diag (τy,1,τy,2,…,τy,p), for the weight coefficient of vehicle yaw stability control, Γr=
diag(τr,1,τr,2,…,τr,p) it is the weight coefficient that vehicle roll stability controls, work as ΓyWhen relatively large, control system side
The control to weaving is overweighted, i.e., practical yaw velocity has preferable tracking effect to its reference value, works as ΓrIt is relatively large
When, control system lays particular emphasis on the Roll control to vehicle, i.e., the size of the reduction lateral load rate of transform as possible.
Γ in object functionyAnd ΓrThe weight that control system controls weaving and roll motion is represented respectively,
The two has opposite meaning, therefore in order to realize variable weight control to adapt to the different stability energy demand of vehicle, herein
We fix ΓyValue, change ΓrValue.Since the lateral load rate of transform of vehicle is the change of direct characterization vehicle roll state
Amount, and steering wheel angle is to directly result in the input quantity that vehicle is turned on one's side, therefore the two parametric variables are to automobile roll shape
The influence of state is maximum.So weight coefficient Γ for Roll controlrDesign one has two inputs and the Fuzzy Control singly exported
Device processed, wherein input quantity are lateral load rate of transform LTR and steering wheel angle δ, and output quantity is Roll control weight coefficient Γr。
Since the degree of danger that vehicle is turned on one's side is more than the degree of danger breakked away, so when the rollover of vehicle is dangerous gradually
When increase, Roll control weight coefficient should also increase therewith.With the aggravation that driver turns to, the LTR of vehicle also exists
Gradual increase, this also gradually increases the danger of vehicle side turning, and Roll control is also more and more important in integrated control, so
ΓrIt should be gradually increased with the increase of δ and LTR.Illustrate herein, when LTR is smaller, vehicle rollover is dangerous
Degree is relatively low, ΓrIt should be slowly increased with the increase of δ;When LTR is larger, vehicle rollover degree of danger steeply rises,
Therefore ΓrSize should aggravate with the increase of δ.
Herein, LTR is blurred, is divided into 3 grades:S (low), M (in), B (height);δ is obscured and is divided into 3 grades:S (low), M
(in), B (height);By ΓrIt is fuzzy to be divided into 4 grades:S (low), M (in), MB (higher), B (height), membership function such as Fig. 2-4
It is shown.It is analyzed according to above-mentioned theory and emulates experience, final design goes out weight coefficient ΓrSpecific fuzzy control about LTR and δ
Rule, as shown in table 1.
1 fuzzy control rule of table
Input/output variable is blurred using Mamdani minimax rationalistic methods and anti fuzzy method operation, finally
Obtain weight coefficient ΓrAbout the three-dimensional map curved surfaces of input quantity LTR and δ, as shown in Figure 5.By above-mentioned fuzzy control strategy,
The design that most weight coefficient is optimized at last.
For the stable constraint problem of vehicle, side slip angle should be made firstPhase plane movement locus is protected
It holds and is stablizing in phase region.That is vehiclePhase plane movement locus should meet the inequality constraints shown in formula (21):
Wherein, k, b are stable constraint constant.
Secondly consider the practical yaw velocity r of vehicle and reference yaw velocity rrefDeviation delta r, lateral load transfer
The size of rate LTR should be limited within stable threshold, therefore there are the stable constraints shown in formula:
Finally consider the saturated characteristic of Vehicular system mechanical structure, control input should meet safety shown in following formula about
Beam:
Wherein, i={ fl, fr, rl, rr }, Δ FxiThe brake force of four tires is represented respectively.
Based on above analysis, most above-mentioned intact stability integrated control problem is described as following optimization problem at last:
4. the tire desired braking power that controller in step 3 is exported is compared with brake pipe actual pressure, so as to decision
Go out to brake the supercharging, pressurize or decompression state of execution system, and be converted into corresponding solenoid valve action command, while will step
The additional front wheel angle exported in rapid 3 is superimposed with the go to action of driver, obtains final front wheel angle;
Optimization problem as shown in formula (24) is solved by non-linear solution tool, and by first item, that is, u of optimized variable
(k) as controller output quantity.Tire brake force Δ FxiIt is expected pressure firstly the need of corresponding wheel cylinder is converted to by formula (25)
Power Pdi:
In above formula:I={ fl, fr, rl, rr }.PdiFor corresponding wheel cylinder brake pressure.DwThe diameter of wheel cylinder is represented, C is represented
Braking effectiveness factor, RbThe radius of brake is represented, n represents oil cylinder number, rdRepresent the rolling radius of tire.
By taking single wheel as an example, the wheel cylinder desired braking pressure P that will obtaini_refWith wheel cylinder actual pressure PiMaking deviation can obtain
ΔPi=Pi-Pi_ref.In control, according to the deviation Δ P of brake pressurei, boosted, protected for the pressure of each wheel cylinder
The adjusting of three kinds of states of pressure and decompression.Frequently switching in order to avoid solenoid valve herein, set the threshold value of pressure divergence as
ΔPth, only when deviation exceeds this threshold value, just boosted or Decompression Controlling, be converted into the transmission of solenoid valve action command
To executing agency.Specific judgement situation is as follows:
(1) as-Δ Pth< Δs Pi< Δs PthPressurize is adjusted
(2) as Δ Pi> Δs PthDecompression is adjusted
(3) as Δ Pi<-Δ PthBoost pressure regulation
Additional front wheel angle is superimposed with driver's go to action, final synthesis obtains front wheel angle δfFor δfd+Δδf。
5th, brake actuator and the solenoid valve action command and front wheel angle that turn in actuator execution step 4 instruct, and make
Vehicle restores stability:
The front wheel angle signal that decision in step 4 goes out is applied to and turns to executing agency, while by the solenoid valve in step 4
Action command is applied to brake actuator, after the completion of being performed by executing agency, vehicle is made to restore to stablize traveling.
Using vehicle dynamics simulation software veDYNA, light auto models are selected, under simulink simulated environment,
The controller model for realizing above-mentioned control program is built, emulation experiment has been done under limit open loop operating mode:It is 0.8 in friction coefficient
Road on, vehicle accelerates to 80km/h, gives steering wheel angle signal as shown in Figure 6, and simulation time is 35 seconds.It is emulated
As a result as illustrated in figures 7 to 13.Find out from simulation result, under the driving cycles of the limit, the intact stability collection designed by the present invention
It can make the ideal yaw velocity of the vehicle upper reference value of tracking, and the lateral load rate of transform also can well into control method
It is fluctuated in stable threshold range, while meets front wheel angle and tire brake force within restriction range, meet vehicle
Yaw stability and roll stability requirement.