DE102007015066B4 - Method and device for regulating drive slip - Google Patents

Abstract

Method for controlling the drive slip, whereby at least the wheel speeds (ω, ω) of the driven wheels corresponding signals are fed to an evaluation circuit, which monitors these wheel speeds (ω, ω) with regard to the occurrence of a tendency to spin and which control signals for braking at least one wheel and / or generated for influencing an engine torque when there is a tendency to spin, and a sum wheel speed variable (ω), which is determined at least from the wheel speed (ω, ω) of the driven wheels, is used for regulation, with the help of a Kalman filter (8) an estimate of the Sum wheel speed variable (ω) is carried out, characterized in that when estimating the sum wheel speed variable (ω) by the Kalman filter, at least one measure for a measurement error (R) is determined and when estimating the sum wheel speed variable (ω), which from the measured wheel speeds (ω, ω), and in particular from the measured engine speed (n), is determined, for the measured wheel speeds (ω, ω), in particular for the sum of the measured wheel speeds (y), and / or for the measured engine speed (n), in particular for the quotient (y) of measured engine speed (n) and an engine speed-wheel speed gear ratio (i), a measure for the measurement error (R) with the aid of two low-pass filters (12, 13), in particular the first Okay, is determined.

Description

The invention relates to a method according to the preamble of claim 1 and a device according to the preamble of claim 9.

From the DE 38 21 769 A1 is a traction control system, in which a quotient of the engine speed and the average wheel speed of the driven wheels is formed and used to control the traction.

Another traction control system for limiting wheel slip during the transmission of a drive torque to the drive wheels is from the US 5 213 177 A known.

Furthermore, apparently the DE 34 46 002 A1 a method for determining the target braking torque for the various wheels of a vehicle, in which signals dependent on the wheel speeds and the vehicle speed are used for this purpose.

It is known that wheel vibrations can lead to a significant oscillation of the controlled variables of a traction control system, which are formed from the signals of the wheel rotation speeds. The filters used to dampen the vibrations lead to a strong phase delay in the controlled variables, i.e. the damped signal lags behind the measurement signal. This leads to a noticeable deterioration in the control quality of the traction control system.

With traction control or traction control, in addition to the speed of the individual wheels, the total speed is also controlled across all drive wheels. The total speed is particularly important in the start-up area in order to keep the engine speed in its optimal speed range.

Malfunctions in the controlled variables due to vibrations of the drive train are particularly problematic. The control according to the total speed is made particularly difficult by drive train vibrations. The usual low-pass filtering of the measured total speed with a first-order filter leads to a significantly poorer signal quality and thus to poorer regulation.

The object of the invention is therefore to provide an improved method for traction control and an improved traction control system.

This object is achieved by the method according to claim 1 and the device according to claim 9.

The invention is based on the idea of carrying out an estimate using a Kalman filter for the controlled variable “total speed of the driven wheels”.

According to the invention, the term “Kalman filter” is understood to mean a stochastic state estimator for dynamic systems.

According to the invention, the term “total speed” is understood in general to mean a quantity that is directly proportional to the total speed. The term thus includes e.g. also half the total speed.

According to the invention, at least one measure for a measurement error is determined and taken into account when estimating the sum wheel rotation speed variable (sum speed), which is determined from the measured wheel rotation speeds, wherein a measure for the measurement error is also determined with the aid of two low-pass filters.

The measure for the measurement error is particularly preferably also determined from the measured engine speed. A measure for a measurement error is very particularly preferably determined from the measured wheel rotational speeds and the measured engine speed in order to take into account the possibly different measurement errors of the individual measurement variables. A measure is advantageously determined from the sum of the measured wheel speeds and a measure from the quotient of the measured engine speed and an engine speed-wheel speed transmission ratio.

First-order low-pass filters are particularly preferred. This method for determining a measure of the measurement error can be carried out both for the measured wheel rotational speeds and for the sum of the measured wheel rotational speeds as well as for the measured engine speed as well as for the quotient of the measured engine speed and an engine speed-wheel speed transmission ratio.

The total speed is preferably estimated on the basis of a model from the measured data of the wheel rotational speeds of the driven wheels and the measured data of the engine speed. For this purpose, the sum of the measured wheel rotational speeds of the driven wheels and the quotient of the measured engine speed and an engine speed-wheel speed transmission ratio are particularly preferably used.

The wheel speed information of the wheel speed sensors of an anti-lock braking system (ABS) is preferably used.

The measure for the measurement error for the measured engine speed is preferably only determined from the measured engine speed when the drive train is closed. Otherwise, with the drive train open, the dimension is set to a predetermined limit. This ensures that the engine speed, which has no connection with the total speed when the drive train is open, has no or only a minor influence in the Kalman filtering. The said applies particularly preferably also to the quotient of the measured engine speed and engine speed-wheel speed gear ratio.

In addition, at least one measure of a model error is preferably taken into account in the estimation of the sum wheel rotational speed magnitude by the Kalman filter, which is selected such that the estimated sum speed is essentially undisturbed, i.e. with slight fluctuations due to the drive train vibrations, and on the other hand essentially without delay. The model error thus represents a type of setting parameter that can be used to adapt to the specific system under investigation.

According to a preferred embodiment of the method according to the invention, the Kalman filter used is divided into two steps. In a first step, among other things, the sum wheel rotational speed size is estimated and a measure for a model error is determined and used. In a second step, among other things, the estimated sum wheel rotational speed variable is improved on the basis of the measured wheel rotational speeds, and particularly preferably on the basis of the measured engine speed, and a measure for a measurement error is determined and used. The variables determined in the second step are then used as output variables for the next first step, etc.

In addition, it is preferred that, with the drive train open, the measure for the model error is set to a predetermined limit value and / or the first step of the Kalman filter is not carried out. This takes into account the fact that the underlying model is only valid when the drive train is closed.

At least one, in particular all, of the following variables is / are preferably taken into account for estimating the sum wheel rotational speed variable: a wheel-related engine torque, a vehicle mass, a vehicle acceleration, a driving resistance, an effective tire radius, an inertia torque of the drive train.

An advantage of the invention is that with the aid of the Kalman filter used, sufficient damping of the vibrations of the controlled variable total speed can be achieved with a significantly smaller phase delay in comparison to the low-pass filtering of the measured total speed known from the prior art. This improves traction control.

Further preferred embodiments result from the subclaims and the following description with reference to figures.

• 1 eine schematische Darstellung eines Kalman-Algorithmus,
• 2 eine schematische Darstellung einer beispielsgemäßen Verfahrens mit Bestimmung einer Summengeschwindigkeit anhand eines Kalman-Filters, und
• 3 eine schematische Darstellung eines Flussdiagramms zur beispielsgemäßen Bildung einer Varianz eines Messfehlers.

Show it

• 1 a schematic representation of a Kalman algorithm,
• 2nd a schematic representation of an exemplary method with determination of a total speed using a Kalman filter, and
• 3rd is a schematic representation of a flowchart for the exemplary formation of a variance of a measurement error.

The following description is an example of a front-wheel drive vehicle. However, the method according to the invention can easily be transferred to rear-wheel or all-wheel drive vehicles. First the Kalman algorithm is briefly explained, then the specific implementation for the method under consideration is discussed.

$$\text{y}\left(\text{k}\right)=\text{C*x}\left(\text{k}\right)+\text{v}\left(\text{k}\right)$$

The Kalman algorithm estimates the state x (k) of a disturbed system in a time-discrete form: $$\text{x}\left(\text{k}\right)=\text{A * x}\left(\text{k}-\text{1}\right)+\text{B}*\text{u}\left(\text{k}\right)+\text{w}\left(\text{k}\right)$$

$$\text{y}\left(\text{k}\right)=\text{C * x}\left(\text{k}\right)+\text{v}\left(\text{k}\right)$$

$$\text{R}=\text{E}\left\{\mathrm{\nu }\left(\text{k}\right){\mathrm{\nu }}^{\text{T}}\left(\text{k}\right)\right\},$$

kann der in 1 schematisch dargestellte Kalman-Algorithmus durchgeführt werden. Der Kalman-Algorithmus wird in einen so genannten „Prädiktionsschritt“ 1 und einen „Innovationsschritt“ 4 unterteilt.Here x (k) describes the state x of the system at a time t _k (abbreviated to k) and x (k-1) describes the state x at a time t _k-1 = t _k -δt earlier by a discrete time interval δt (abbreviated to k-1). The underlying model, for example to describe the time development of the system under consideration (vehicle), is described by the operators / matrices A, B, and C and w (k) or v (k) describe the error of the model used or a Measurement error at time t _k . y (k) describes measurement data at time t _k and u (k) describes the input variables of the system (eg influenced by the driver, such as wheel brake pressure or engine torque). Given the model error covariance matrix (process noise) Q, which results as the expected value E from the model error w (k), and the given measurement error covariance matrix R, which results as the expected value E from the measurement error v (k): $$\text{Q}=\text{E}\left\{\text{w}\left(\text{k}\right){\text{w}}^{\text{T}}\left(\text{k}\right)\right\}$$

$$\text{R}=\text{E}\left\{\mathrm{\nu }\left(\text{k}\right){\mathrm{\nu }}^{\text{T}}\left(\text{k}\right)\right\},$$

can the in 1 schematically represented Kalman algorithm can be performed. The Kalman algorithm is divided into a so-called “prediction step” 1 and an “innovation step” 4.

When "prediction" 1 is out of the previous state x (k-1) at time t _k-1 by using the state space model (described by A, B and C), the new state x ^- (k) at time t _k predicted (block 2nd : Estimate of new conditions): $${\text{x}}^{-}\left(\text{k}\right)=\text{A}*\text{x}\left(\text{k}-1\right)+\text{B}*\text{u}\left(\text{k}\right)$$

State x ^- (k) is only an estimate of state x (k). The error covariance P ^- (k) of this estimate is determined from the old error covariance P ^- (k-1) and the process noise Q (block 3rd : Update error covariance), while the system matrix A and its transpose A ^T is also a: $${\text{P}}^{-}\left(\text{k}\right)=\text{A}*{\text{P}}^{-}\left(\text{k}-1\right)*{\text{A}}^{\text{T}}+\text{Q}$$

In the "innovation step" 4th the Kalman gain K is determined from the error covariance P ^- (k), the system matrix C and the measurement error covariance matrix R (block 5 : Calculation of the Kalman gain), where a superscript T stands for the transpose of a matrix / vector and a superscript index -1 for the inverse of a matrix: $$\text{K}\left(\text{k}\right)={\text{P}}^{-}\left(\text{k}\right)*{\text{C.}}^{\text{T}}*{\left({\text{CP}}^{-}\left(\text{k}\right){\text{C.}}^{\text{T}}+\text{R}\right)}^{-1}$$

An improved (ie a posteriori) estimated value x (k) is determined from the previously estimated (ie a priori) state x ^- (k) and a new measurement point y (k) by a linear combination with the Kalman gain K (block 6 : Correction of the estimate using the measurement data): $$\text{x}\left(\text{k}\right)={\text{x}}^{-}\left(\text{k}\right)+\text{K}\left(\text{k}\right)*\left(\text{y}\left(\text{k}\right)-\text{C.}*{\text{x}}^{-}\left(\text{k}\right)\right)$$

In the last step (block 7 : Update error covariance matrix) the new error covariance P (k) is determined from the unit matrix I, the Kalman gain K, the system matrix C and the error covariance P ^- (k): $$\text{P}\left(\text{k}\right)=\left(\text{I.}-\text{K}\left(\text{k}\right)*\text{C.}\right)*{\text{P}}^{-}\left(\text{k}\right)$$

• • Die Messdaten y(k)
• • Das diskrete Prozessmodell (beschrieben durch A, B und C)
• • Die Messfehlerkovarianzmatrix (Varianzen des Messrauschens) R
• • Die Modellfehlerkovarianzmatrix Q

To use the Kalman algorithm according to 1 The following information must therefore be known:

• • The measurement data y (k)
• • The discrete process model (described by A, B and C)
• • The measurement error covariance matrix (variances of the measurement noise) R
• • The model error covariance matrix Q

According to the invention, the Kalman algorithm is used to estimate the total speed of the drive wheels. According to the example, the total speed is estimated by Kalman filtering from the measurement data of the wheel speeds and the measurement data of the engine speed. This means that the undisturbed wheel speeds are still unknown.

However, by adapting the traction control model, the total speed can be replaced by the most disturbed wheel speed. The sensitivity of the control system to vibrations of the drive train can thus be reduced.

When using a Kalman filter according to the invention for estimating the total velocity, the system state x of the Kalman algorithm corresponds to the sought overall velocity ω _s . In 2nd is a schematic of an exemplary method with determination of a total speed ω _s using a Kalman filter 8th shown. The determined total speed ω _s becomes the traction control system 9 used. For the system state x (k), two measured values y ₁ (k) and y ₂ (k) are available at every point in time k, for example. The measured value y ₁ results from the sum of the wheel rotation speed signals ω _L and ω _{R of} the left and right driven wheels (each at a time k): $${\text{y}}_1\left(\text{k}\right)=\left[{\mathrm{\omega }}_{\text{L}}\left(\text{k}\right)+{\mathrm{\omega }}_{\text{R}}\left(\text{k}\right)\right]/\mathrm{2nd}$$

Marked on front or rear-wheel drive wheels ω _L or ω _{R is} the wheel rotation speed of the left or right drive wheel. In all-wheel drive vehicles with a controllable center clutch, ω _L or ω _{R is} the wheel rotation speed of the left or right drive wheel of the main drive axle.

The measured value y ₂ is determined from the engine speed n by division by an estimated (predetermined) transmission ratio i engine speed to wheel speed: $${\text{y}}_{\mathrm{2nd}}\left(\text{k}\right)=\text{n}\left(\text{k}\right)/\text{i}\left(\text{k}\right)$$

The two measured values are combined to form a measurement vector y: $$\text{y}\left(\text{k}\right)={\left[{\text{y}}_{\text{1}}\left(\text{k}\right){\text{y}}_{\mathrm{2nd}}\left(\text{k}\right)\right]}^{\text{T}}$$

angeben. Das Störsignal v(k) in der Messgleichung (2) beinhaltet neben den Messfehlern auch die oben diskutierten Antriebsstrangschwingungen. Es setzt sich aus der Störung v₁(k) als Störung von y₁(k) und v₂(k) als Störung von y₂(k) zusammen: $$\mathrm{\nu }\left(\text{k}\right)={\left[\begin{array}{cc}{\mathrm{\nu }}_1\left(\text{k}\right)& {\mathrm{\nu }}_2\left(\text{k}\right)\end{array}\right]}^{\text{T}}$$

The output matrix C can thus be used in the measurement equation (2) $$\text{C.}={\left[\begin{array}{cc}1& 1\end{array}\right]}^{\text{T}}$$

specify. In addition to the measurement errors, the interference signal v (k) in the measurement equation (2) also includes the drive train vibrations discussed above. It consists of the disturbance v ₁ (k) as disturbance of y ₁ (k) and v ₂ (k) as disturbance of y ₂ (k): $$\mathrm{\nu }\left(\text{k}\right)={\left[\begin{array}{cc}{\mathrm{\nu }}_1\left(\text{k}\right)& {\mathrm{\nu }}_{\mathrm{2nd}}\left(\text{k}\right)\end{array}\right]}^{\text{T}}$$

Assuming that the disturbances v ₁ and v _{2 are} white (ie, for example, they are the same at all frequencies), mean-free and uncorrelated to one another, the covariance matrix R can be determined from the variances of the two disturbances: $$\text{R}=\text{E}\left\{\mathrm{\nu }{\mathrm{\nu }}^{\text{T}}\right\}=\text{E}\left\{\begin{array}{cc}\mathrm{{\rm E}}\left\{{\mathrm{<figure-callout\; id="1"\; label="\nu "\; filenames="DE102007015066B4\_0021.png"\; state="\{\{state\}\}">\nu </figure-callout>}}_1{\nu }_1{}^{\text{T}}\right\}& 0\\ 0& \mathrm{{\rm E}}\left\{{\nu }_{\mathrm{2nd}}{\nu }_{\mathrm{2nd}}{}^{\text{T}}\right\}\end{array}\right\}$$

$$\text{E}\left\{\begin{array}{cc}{\mathrm{\nu }}_2& {\mathrm{\nu }}_2{}^{\text{T}}\end{array}\right\}\approx \text{E}\left\{\begin{array}{cc}{\text{y}}_2& {\text{y}}_2{}^{\text{T}}\end{array}\right\}$$

To estimate the strongly time-variant measurement error covariance matrix R, their proportions E {v ₁ v ₁ ^T } and E {v ₂ v ₂ ^T } are approximated using the variances of the measured values y ₁ and y ₂ , for example: $$\text{E}\left\{\begin{array}{cc}{\mathrm{<figure-callout\; id="1"\; label="\nu "\; filenames="DE102007015066B4\_0021.png"\; state="\{\{state\}\}">\nu </figure-callout>}}_1& {\mathrm{\nu }}_1{}^{\text{T}}\end{array}\right\}\approx \text{E}\left\{\begin{array}{cc}{\text{y}}_1& {\text{y}}_1{}^{\text{T}}\end{array}\right\}$$

$$\text{E}\left\{\begin{array}{cc}{\mathrm{\nu }}_{\mathrm{2nd}}& {\mathrm{\nu }}_{\mathrm{2nd}}{}^{\text{T}}\end{array}\right\}\approx \text{E}\left\{\begin{array}{cc}{\text{y}}_{\mathrm{2nd}}& {\text{y}}_{\mathrm{2nd}}{}^{\text{T}}\end{array}\right\}$$

3rd shows a schematic representation of a flow chart for the exemplary formation of a variance of a measurement error. The variances of y ₁ (i = 1 or 2) are determined by two first-order low-pass filters at runtime 12th , 13 (PT1 filter) according to 3rd estimated. First filtering 12th the mean value E {y _i } of the measured value y _i results. Through addition (subtraction) 14 of measured value y ₁ and mean value E {y _i }, subsequent multiplication 15 the result with itself and the following second filtering 13 the variance E {y _i -E {y _i }} ² results.

For example, the variance E {y ₁ y ₁ ^T } is always in accordance 3rd formed, while the variance E {y ₂ y ₂ ^T } only according to when the drive train is closed 3rd is determined. In the case of an open drive train, there is no relationship between engine speed n and total speed. The measured value y ₂ (k) is therefore no longer reliable and the value of E {y ₂ y ₂ ^T } is set to a value much larger than E {y ₁ y ₁ ^T }. For example, the value of E {y ₂ y ₂ ^T } can be set to a maximum permissible value. This eliminates or reduces the influence of engine speed n (or y ₂ ) on the Kalman filter. A closed drive train is always suspected when a wheel shows traction.

wobei M_E [Nm] das radbezogene Motormoment, M_B,L [Nm] das Bremsmoment des linken Rades, M_B,R [Nm] das Bremsmoment des rechten Rades, m [kg] die Masse des Fahrzeugs, a [m/s²] die Fahrzeugbeschleunigung, p [m/s²] der Fahrwiderstand, reff [m] der effektive Reifenradius, i das Übersetzungsverhältnis, J_D [kg*m/s²] die antriebsseitige und J_DA [kg*m/s²] die abtriebsseitige Trägheitsmasse des Differentials ist. Daraus lassen sich nach Zeitdiskretisierung (t_sample sei die Abtastzeit) und Vergleich mit Gleichung (1) die noch unbekannten Systemmatrizen A und B sowie der Eingang u angeben: $$\begin{array}{l}\text{A}=1\\ \text{B}={\text{t}}_{\text{sample}}\\ \text{u}\left(\text{k}\right)=\left[{\text{M}}_{\text{E}}\left(\text{k}\right)-{\text{M}}_{\text{B},\text{L}}\left(\text{k}\right)-{\text{M}}_{\text{B},\text{R}}\left(\text{k}\right)-\text{m}*\left(\text{a}\left(\text{k}\right)+\text{p}\left(\text{k}\right)\right)*{\text{r}}_{\text{eff}}\right]/\left[2{\text{J}}_{\text{DA}}+{\text{i}}^2*{\text{J}}_{\text{D}}\right]\end{array}$$

In accordance with the example, the following time-continuous relationship with respect to the time derivative of the total velocity dω _s / dt (corresponds to dx (t) / dt) is used to determine the system matrices A and B. $$\begin{array}{l}\text{dx}\left(\text{t}\right)/\text{German}=\\ \left[{\text{M}}_{\text{E}}\left(\text{t}\right)-{\text{M}}_{\text{B},\text{L}}\left(\text{t}\right)-{\text{M}}_{\text{B},\text{R}}\left(\text{t}\right)-\text{m *}\left(\text{a}\left(\text{t}\right)+\text{p}\left(\text{t}\right)\right)*{\text{r}}_{\text{eff}}\right]/\left[\mathrm{2nd}{\text{J}}_{\text{THERE}}+{\text{i}}^{\text{2nd}}*{\text{J}}_{\text{D}}\right]\end{array}$$

where M _E [Nm] the wheel-related engine torque, M _{B, L} [Nm] the braking torque of the left wheel, M _{B, R} [Nm] the braking torque of the right wheel, m [kg] the mass of the vehicle, a [m / s ² ] the vehicle acceleration, p [m / s ² ] the driving resistance, reff [m] the effective tire radius, i the gear ratio, J _D [kg * m / s ² ] the drive side and J _DA [kg * m / s ² ] is the inertial mass of the differential on the output side. The system matrices A and B, as well as the input u, can be specified from this after time discretization (t _sample is the sampling time) and comparison with equation (1): $$\begin{array}{l}\text{A}=1\\ \text{B}={\text{t}}_{\text{sample}}\\ \text{u}\left(\text{k}\right)=\left[{\text{M}}_{\text{E}}\left(\text{k}\right)-{\text{M}}_{\text{B},\text{L}}\left(\text{k}\right)-{\text{M}}_{\text{B},\text{R}}\left(\text{k}\right)-\text{m}*\left(\text{a}\left(\text{k}\right)+\text{p}\left(\text{k}\right)\right)*{\text{r}}_{\text{eff}}\right]/\left[\mathrm{2nd}{\text{J}}_{\text{THERE}}+{\text{i}}^{\mathrm{2nd}}*{\text{J}}_{\text{D}}\right]\end{array}$$

mit einem Tiefpassfilter erster Ordnung abgeschätzt werden.At the input u, all sizes except the driving resistance p (k) can be measured or specified. The driving resistance p can be achieved by filtering the signal

can be estimated with a first order low pass filter.

According to the example, the model error covariance matrix Q of the disturbance w (k) is used as a tuning factor in the case of the closed drive train in order to achieve a good compromise between damping the disturbances and a sufficiently fast subsequent behavior to the measurement data. For this purpose, the elements of the model error covariance matrix Q are varied as "free parameters" in a certain interval. The model (equations (3) and (4)) is not valid for an open drive train. In this case, Q is set to a value Q = E {w (k) w ^T (k)} much larger than E {y ₁ (k) y ₁ ^T (k)} and the prediction step 1 from the Kalman filter 8th hidden.

According to the prior art, it is known to carry out low-pass filtering of the sum signal of the measured wheel speeds (of the drive wheels) in order to determine the total speed. This corresponds to a simple low-pass filtering of the measured values y ₁ . Through the Kalman filter 8th compared to this low-pass filtering of the sum signal of the measured wheel speeds, an improved separation of useful and interference signals is achieved. This applies both to the attenuation of the interference and to the phase delay with respect to the original useful signal.

Claims (10)

Method for controlling the drive slip, wherein at least the wheel rotation speeds (ω _R , ω _L ) of the driven wheels are supplied with signals corresponding to an evaluation circuit which monitors these wheel rotation speeds (ω _R , ω _L ) for the occurrence of a tendency to spin and which control signals for braking at least one Wheel and / or for influencing an engine torque in the event of spinning, and wherein a sum wheel rotation speed variable (ω _s ), which is determined at least from the wheel rotation speeds (ω _R , ω _L ) of the driven wheels, is used for control, with the aid of a Kalman Filters (8) an estimate of the sum wheel rotation speed quantity (ω _s ) is performed, characterized in that when the sum wheel rotation speed quantity (ω _s ) is estimated by the Kalman filter, at least one measure for a measurement error (R) is determined and in the estimation of the sum wheel rotation speed quantity ( ω _s ) is taken into account, which is determined from the measured wheel rotational speeds (ω _R , ω _L ), and in particular from the measured engine speed (n), whereby for the measured wheel rotational speeds (ω _R , ω _L ), in particular for the sum of the measured wheel rotational speeds ( y ₁ ), and / or for the measured engine speed (n), in particular for the quotient (y ₂ ) from the measured engine speed (n) and an engine speed-wheel speed transmission ratio (i), each with a measure of the measurement error (R) With the help of two low-pass filters (12, 13), in particular first order, is determined. Procedure according to Claim 1 , characterized in that for estimating the sum wheel rotational speed variable (ω _s ) at least measured wheel rotational speeds (ω _R , (ω _L ) of the driven wheels, in particular a sum of the measured wheel rotational speeds (y ₁ ) of the driven wheels, and a measured engine speed (n), in particular a quotient (y ₂ ) from the measured engine speed (n) and an engine speed-wheel speed transmission ratio (i) can be used. Procedure according to Claim 1 or 2nd , characterized in that only then the measure for the measurement error (R) for the measured engine speed (n), in particular for the quotient (y ₂ ) from the measured engine speed (n) and an engine speed-wheel speed transmission ratio (i), from the measured engine speed (n) is determined when the drive train is closed. Method according to at least one of the preceding claims, characterized in that the measure for the measurement error (R) for the measured engine speed (n), in particular for the quotient (y ₂ ) from the measured engine speed (n) and an engine speed-wheel speed transmission ratio ( i) is set to a predetermined limit when the drive train is open. Method according to at least one of the preceding claims, characterized in that at least one measure for a model error (Q) is taken into account in the estimation of the sum wheel rotation speed quantity (ω _s ) by the Kalman filter, which is selected such that the estimated sum wheel rotation speed quantity (cos) essentially undisturbed and without delay in relation to the measured quantities. Method according to at least one of the preceding claims, characterized in that when the sum wheel rotational speed variable (ω _s ) is estimated by the Kalman filter, a first step (1) in which the sum wheel rotational speed variable (ω _s ) is estimated and a measure for a model error ( Q) is used, and a second step (4), in which the estimated sum wheel rotational speed magnitude (ω _s ) is improved at least on the basis of the measured wheel rotational speeds (ω _R , ω _L ), and in particular on the basis of the measured engine speed (n), and a measure is used for a measurement error (R). Procedure according to Claim 5 or 6 , characterized in that the measure for the model error (Q) is set to a predetermined limit value and / or the first step (1) of the Kalman filter (8) is not carried out when the drive train is open. Method according to at least one of the preceding claims, characterized in that at least one, in particular all, of the following variables is / are taken into account for estimating the sum wheel rotational speed variable (ω _s ): a wheel-related engine torque (M _E , M _{B, L} , M _{B, R} ), a vehicle mass (m), a vehicle acceleration (a), a driving resistance (p), an effective tire radius (r _eff ), an moment of inertia (J _{D, A} , J _D ) of the drive train. Device for controlling the drive slip, wherein at least the signals corresponding to the wheel rotation speeds (ω _R , ω _L ) of the driven wheels are fed to an evaluation circuit which monitors these wheel rotation speeds (ω _R , ω _L ) with regard to the occurrence of a tendency to spin and which control signals for braking at least one Generated wheel and / or to influence an engine torque when spinning, and which includes a means for determining a sum wheel rotation speed quantity (ω _s ) from the wheel rotation speeds (ω _R , ω _L ) of the driven wheels, the sum wheel rotation speed quantity (ω _s ) used for slip control characterized in that it comprises a Kalman filter (8) with which an estimate of the sum wheel rotation speed variable (ω _s ) is carried out, with the Kalman filter estimating the sum wheel rotation speed variable (cos) at least one measure for a measurement error ( R) is determined and taken into account in the estimation of the sum wheel rotational speed variable (ω _s ), which is determined from the measured wheel rotational speeds (ω _R , (ω _L ), and in particular from the measured engine speed (n)), with the measured wheel rotational speeds (ω _R , ω _L ), in particular for the sum of the measured wheel speeds (y ₁ ), and / or for the measured engine speed (n), in particular for the quotient (y ₂ ) from the measured engine speed (n) and an engine speed-wheel speed gear ratio (i ), a measure of the measurement error (R) is determined with the help of two low-pass filters (12, 13), in particular of the first order. Device after Claim 9 , characterized in that in this a method according to one of the Claims 1 to 8th is carried out.

Images