DE4221030C2 - Method for recognizing the situation of a vehicle - Google Patents

Description

In recent years, vehicles with an active rear axle steering offered on the market. From the beginning pure control concepts in actively influencing the back Axle kinematics can be a trend in the direction of regulated concepts are observed in the development of the chassis. Have both simulations and experimental results shown that even more intelligent strategies to fulfill strict requirements regarding security, stability and the driving comfort of a complex motor vehicle at under different driving maneuvers, especially in borderline cases Need to become. So it was found that for cornering and Lane change completely opposite control measures of the rear axle steering must be used. Therefore, it is desirable to know the current situation. The compensation of the Brake yaw moment by rear axle steering (GMK) is a very sensible full measure with µ-split braking, but brings undesirable Effects when cornering and changing lanes. Therefore, based on Lateral acceleration signals the GMK function in the latter Situations weakened (indirect  Situation adjustment). This shows how important the detection of Driving situations when filling a driving dynamics control can.

DE 40 30 846 A1 describes a method for controlling or Regulation of a rear axle steering angle is known, with Hil The fuzzy logic matches the control behavior to the driving state can be fitted.

DE 38 87 015 T2 describes a method and device for Brake pressure control using fuzzy logic is known, both Fuzzy controller as well as determination of unknown quantities using Fuzzy logic can be described.

The object of the invention is therefore to develop a technically feasible method which makes it possible to recognize the present situation on-line by evaluating a small number of measurement signals available in the vehicle. Certain restrictions / requirements are inevitably stipulated because of the motor vehicle conditions; namely

• - the situation must be recognized quickly,
• - the computing capacity is limited to on-board computers,
• - The error rate of the classification must be kept very low.

In the latter point, it is particularly critical if the available data / features are imprecise (imprecise due to measurement, not clear in the evaluation, etc.). It is therefore the intention to support this recognition problem by using "fuzzy logic". By using fuzzy logic, there are the parts before that

• - easier to handle ambiguous situations (decisions) are,
• - The effort for the evaluation is limited.

With regard to the yaw moment compensation (GMK) project as practical application a method should be described the cornering braked by the driving situation from the driving situation Braking on µ split can differentiate. The result is in Form of a signal that has values between 0. , , 1 can assume to Available. The output signal is continuously updated so that also changes in the driving situation during braking be taken into account. The result is at the latest 100 msec Braking begins or after a change in the driving situation occurs to disposal.

The following signals, which are also required for GMK and ABS, are available:

• - steering angle front and rear (δ _v , δ _h ),
• - Front left and front right brake pressures (P _vl , P _vr )
• - Vehicle speed (v _x ).

Based on the drawing, exemplary embodiments of the invention are intended are explained.

Show it. Fig. 1 a control loop using different controls, Fig. 2 is a block diagram of a classifier, Figs. 3 to 6 are diagrams, Fig. 7 is a detailed embodiment of a classifier.

In Fig. 1, 1 is a vehicle on which certain quantities are measured. In a classifier 2 , the situation µ-split braking or corner braking or straight braking is recognized and, depending on the situation, a switch is made to one of three controllers 3 with different control sets. B. optimally adjusts the rear axle steering angle δ _{h of} a vehicle in the sense of compensation of the yaw moment.

Fig. 2 shows a block diagram of the classifier 2 consisting of the blocks feature generator 2.1 , assessment 2.2 and Situationszu folder 2.3 .

The signals from the sensors available in the vehicle (δ _v , δ _h , P _vl , P _vr , V _x ) are fed to the feature generator at a terminal 4 . From this, a suitable feature vector x _M (sj) is formed, which characterizes the corresponding situations on the basis of automotive engineering / control engineering features. This feature vector will

• - be easy to win,
• - sufficient for further processing in the classifier Provide information

and

• - As few ambiguities as possible for the decision hold.

The feature vector serves as the basis for the situation recognition. With the help of the evaluator 2.2 , which works on the basis of the fuzzy logic, the feature vector is evaluated. The result of the evaluation is used further by a situation assignor to make a decision as to which situation is present.

One approach for the feature creator is the combination of the following sizes:

• - lateral acceleration b _y , the z. B. can be derived from δ _v , δ _h and v _x
• - brake pressure difference ΔP = P _vl - P _{vr '}

and

• - Brake pressure difference ΔP _mod in the event of a corner braking.

The stationary lateral acceleration can be calculated from the steering angle at the front and rear and from the vehicle speed using the following relationship

The characteristic speed v _ch is a parameter dependent on vehicle data. This can even have a characteristic speed which is dependent on the estimated lateral acceleration. l _o is the center distance.

Since the vehicle cannot follow the steering angle directly, Simulation of the driving dynamics a dynamic link (PT1) turned on on. The driving dynamics also change with the cross acceleration. If the values are small, the vehicle reacts faster on steering angle changes than for large ones.

In order to differentiate between the different driving situations, the previous history is also important in some cases. For this reason, the lateral acceleration is estimated according to the above equation from the very heavily filtered front steering angle. A moving averaging over a fairly long period of time is used as the steering angle filter. In this case, no large lateral accelerations are estimated for short-term, fast dynamic steering interventions, whereas, for longer turns, the lateral accelerations estimated with filtered and unfiltered steering angles are almost the same. This estimate is carried out in block 5 in the detailed exemplary embodiment in FIG. 7.

The pressure difference between the left and right wheels is a measure of the difference in coefficient of friction and thus for the torque around the vertical axis of the vehicle:

ΔP = P _vl - P _vr

This difference is formed in a block 6 .

We are looking for the transfer function between the steering angle and the brake pressure difference in the event of cornering braking. This modeled brake pressure difference ΔP _mod is to be compared with the measured brake pressure difference ΔP. In the case of corner braking, a good match between the two signals is to be expected, while larger deviations occur in the case of a µ-split braking.

You now define a learning factor

which characterizes the different driving situations as shown in FIG. 3:
α <1 µ split braking
α = 1 corner braking
α <1 straight braking

With the help of this factor you can see whether there is a curve braking or non-curve braking is present.

This factor is "learned" from existing measurement signals. Lateral acceleration is a potential parameter. The learning factor α is therefore simulated as a function of the estimated lateral acceleration, which is obtained at the top from the steering angle at the front, steering angle at the rear and vehicle speed.

α = α (b _y )

In Fig. 4, the context for the vehicle is Darge provides. The situations of corner braking and µ-split braking are described qualitatively here in the form of the factor α (block 7 ).

The case of straight braking is taken into account by introducing the following evaluation variable γ:

It can be seen from FIG. 5 that the very pronounced minimum of the evaluation variable r (= 0) characterizes the braking of the curve (block 8 ). The distinction between the driving situations µ-split braking and straight braking is not clear from this. You still need the quantity ΔP, which is formed in block 6 .

An evaluation variable was thus found without modeling, which qualitatively describes the ratio ΔP _mod / ΔP.

The corresponding membership functions can now be formed from the variables b _y , ΔP and r:

where the parameters w can be selected as follows

w _by = 10

w _{Δ P} = 0.04

w _r = 0.5

This process is done in the evaluator 2.2 in blocks 9 , 10 and 11 before. The associated courses are shown in FIG. 6.

The following rules are used in situation assignor 2.3 for situation detection:
Rule 1.1 IF µ _by BIG THEN corner braking (Block 12 )
Rule 2.1 IF µ _r BIG and µ _{Δ P} BIG THEN µ split braking (block 13 )
Rule 3.1 IF µ _r BIG and µ _{Δ P} SMALL THEN straight braking (Block 14 )

Min or product operators can be used for fuzzy implication (Pedrycz 1989). Eg for rule 2.1 :
min-operator:

µsplit = min (µ _r , µ _{Δ P} )

product-operator:

µsplit = µ _r × µ _{Δ P}

The maximum of all rules, which is formed in block 15 , applies to the final decision about the situation detection.

The following criteria can be introduced to ensure the decision:

Claims (9)

1. Method for recognizing one of the driving situations curves braking or straight braking or µ-split braking, in who has a braked vehicle, where using fuzzy logic and dependent speed of a determined lateral acceleration for the vehicle, a difference in the measured brake pressures of the front wheels and one depending on the lateral acceleration by ma thematic operations formed valuation size that a qualitative measure for the characterization of the driving situation represents the current driving situation recognized becomes.   2. The method according to claim 1, characterized in that that the lateral acceleration of the vehicle depending on the Steering angle for the front wheels, the steering angle for the rear wheels and the vehicle speed is determined. 3. The method according to claim 1, characterized in that depending on the detected driving situation on one of three controllers with different control laws switched becomes. 4. The method according to claim 3, characterized in that with one of the controls the rear axle steering angle of the vehicle optimally in terms of compensation of a yaw moment is posed. 5. The method according to claim 1, characterized in that the vehicle is equipped with an anti-lock braking system is.   6. The method according to claim 1, characterized in that features are initially formed to identify the driving situation that characterize the corresponding situation conclude these features based on fuzzy logic and depending on the result of the evaluation it is decided which driving situation is present. 7. The method according to claim 6, characterized in that the features depending on the steering angle for the front the wheels, the steering angle for the rear wheels, the brake pressure kes of the left front wheel, brake pressure of the right forward derrades and the vehicle speed are formed.   8. A method for detecting a situation in which there is a braked vehicle equipped with an anti-lock braking system, characterized in that the lateral acceleration b _y and the difference in the braking pressures Δ _{P of} the front wheels are determined so that a quantity α is given off from the lateral acceleration b _y conducts, which is 0.5 at small lateral accelerations, increases with increasing lateral acceleration to 1 and then 1 remains that from this in accordance with the relationship
γ = abs (1 / α - 1)
a size r is determined in accordance with the relationships
and
Variables µ _by , µ _{Δ P} and µ _{r are} obtained, whereby w _by , w _{Δ P} and w _{r are} constant, that with µ _by big a curve brake signal µ _r with µ _{Δ P} big and µ _r big a µ-split Brake signal µ _split and at µ _{Δ P} small and µ _r big a straight brake signal µ _{δ is} obtained and that the maximum value of the signals µ _K , µ _Split and µ _{δ is} selected to identify the situation. 9. Device for recognizing one of the driving situations cure braking or straight braking or µ-split braking, in who has a braked vehicle, with that using fuzzy logic and dependent speed of a determined lateral acceleration for the vehicle, a difference in the measured brake pressures of the front wheels and one depending on the lateral acceleration by ma thematic operations formed valuation size that a qualitative measure for the characterization of the driving situation represents the current driving situation recognized becomes.