DE4123053C2 - Method for determining at least one movement variable of a vehicle - Google Patents

Description

State of the art

The yaw rate ω and / or the transverse speed v _y in the vehicle as well as the slip angle α and the transverse forces F _y are difficult or only measurable with expensive sensors. In the older patent application according to 40 30 653, the determination of small slip angles using an expensive yaw sensor is known.

Such sizes play an important role in chassis control.

Advantages of the invention

In the invention, the lateral accelerations of the vehicle in front of and behind the center of gravity and the steering angle of the axles are measured using inexpensive sensors, the rear axle steering angle also being 0 (no rear axle steering). When using the method according to the invention, first of all the transverse speed and the yaw rate arise. From this, the slip angle α, the tire forces F _y and the slip angle β can then be derived in a simple manner.

Figure description

As shown in Fig. 1, the tire dynamics in previous considerations is approximated by a linear equation with constant parameter C _α :

f _y = C _α . α (1)

This approach only applies to very small slip angles. With increasing slip angle, the actual and the lateral force modeled by ( 1 ) no longer match.

It was therefore the object of the invention to first of all provide an approach for the lateral force F _y which covers the entire skew region. With the help of this approach, a technically feasible method is then found, which, from the measured variables δ _v , δ _h , a _yv , a _yh, by a combination of adaptive equivalent Kalman filters, first the yaw rate ω and the lateral speed v _y and finally the slip angle α and the lateral forces F _y determined.

The first part of this task is solved by the approach sketched in Fig. 2, which states that the tire dynamics by a linear approach

F _y (α) = k _y (α). α + h _y (α) (2)

can be described with variable parameters k _y (α) and h _y (α).

A comparison with the approach in ( 1 ) shows that

• - The skew area is completely writable and
• - The tire behavior is described more realistically by different parameters k _y (α) and h _y (α) depending on α.

The difficulty lies in determining these parameters with simple means. This is achieved through the arrangement shown in FIG. 3. If the values of α, k _y (α) and h _y (α) are available, then F _y can easily be calculated from ( 2 ). The arrangement of FIG. 3 is technically feasible; k _y , h _y and α are estimated as follows.

The input variables of the estimator are the measured lateral acceleration a _{yv in} front of the center of gravity, the measured lateral acceleration a _yh behind the center of gravity, the steering angle at the front δ _v and the steering angle at the rear δ _h , which will be required later.

The lateral accelerations are combined once as a sum ( FIG. 3, upper strand) and another time as a difference ( FIG. 3, lower strand) (blocks 1 and 2 ) and so on. The overall structure of the estimator is divided into two subsystems, with the subsystems containing identical structures and differing only in the input variables. Therefore, only one subsystem will be considered below.

The subsystem consists of two adaptive equivalent Kalman filters 3 and 4 , the filter 3 approximating the lateral force according to equation (1) and the second extended filter 4 (E-) describing the lateral force using equation (2). Each filter provides an "estimate of the input variable". By _forming the difference between the two estimated values (difference generator 5 ), a function g (h _yv , h _yh ) in the upper subsystem and a function h (h _yv , h _yh ) in the lower subsystem of h _yv and h _{yh are} obtained. This results in a solvable system of equations with two equations and two unknowns (block 6 ). In addition, the estimate with the filter 4 provides the parameters k _yv (α) and k _yh (α) from the relationships

With the invention, the parameters k _yv , k _yh , h _yv and h _{yh can} be determined explicitly. The state variables transverse speed v _y and yaw rate ω can be reconstructed from the parameter vector of the filter 4 (for a derivation see later). From these two motion quantities can be made according to the following relationship

calculate the slip angle α, so that now all sizes for calculation the lateral forces according to equation (2) are available.

The basic equation for AkF is assumed in filter 3 :

in which

y (k) = ay (k) = a _yv (k) ± a _yh (k)

u ₁ (k) = δ _v (k)

u ₂ (k) = δ _h (k)

are.

The parameter vector

p = [p _y1 p _y2 | p _u11 p _u12 P _u13 | P _u21 P _u22 p _u23 ] ^T

can from the data vector m (k - 1)

m (k - 1) = [y (k - 2) y (k - 3) u ₁ (k - 1) u ₁ (k - 2) u ₁ (k - 3) u ₂ (k - 1) u ₂ (k - 2) u ₂ (k - 3)] ^T

and an error

can be calculated recursively using an estimation algorithm:

where r (k) with "stochastic approximation" z. B. is selected as follows:

The state rector

x = [ω v _y ] ^T

can then be obtained as follows:

where the matrices M. (k - 1) can be read from the data vector m ^T (k - 1) and the parameter vector. (k) from the estimate vector (k):

Filter 4 assumes the basic equation for E-AkF:

in which

y (k) = a _y (k) = a _yv (k) ± a _yh (k),

u ₁ (k) = δ _v (k),

u ₂ (k) = δ _h (k) and

u ₃ (k) = γ (k)

(γ (k) is a freely selectable function).  

The parameter vector

p = [p _y1 p _y2 | p _u11 p _u12 p _u13 | p _u21 p _u22 p _u23 | p _u31 p _u32 p _u33 ] ^T can

from the data vector m (k - 1)

m (k - 1) = [y (k - 2) y (k - 3) | u ₁ (k - 1) u ₁ (k - 2) u ₁ (k - 3) |

u ₂ (k - 1) u ₂ (k - 2) u ₂ (k - 3) | u ₃ (k - 1) u ₃ (k - 2) u ₃ (k - 3)]

and the mistake

can be calculated recursively using an estimation algorithm:

where z. B. r (k) for "stochastic approximation" is selected as follows:

The state rector

x = [ω v _y ] ^T

can then be obtained as follows

where the matrices M. (k - 1) can be read from the data vector m ^T (k - 1) and the parameter vector. (k) from the estimation factor (k):

In addition to ω and v _y, the other motion quantities can be derived from this

Claims (1)

Method for determining at least one movement variable (such as the transverse speed v _y , the yaw rate ω, the _slip angle β, the _slip angle α _v , α _h and the lateral forces F _yv , F _yh ) of a motor vehicle, characterized in that the measured variables lateral acceleration a _yv before and the lateral acceleration a _yh behind the center of gravity are determined so that these lateral accelerations in the form of a sum (a _yv + a _yh ) parallel to two first adaptive equivalent Kalman filters (1st filter pair) and in the form of a difference ( a _yv - a _yh ) two additional adaptive equivalents, the same structure Kalman filters (2nd filter pair) are fed in that a filter of each filter pair (AkF) is designed so that it has the lateral forces F _{y of} the tires according to the relationship
F _y = C _α . α
approximates, where α is the slip angle and C _{α is} a constant that the other extended filter (E-AkF) of each filter pair is designed to have the lateral forces F _{y of} the tires according to the relationship F _y = k _y (α). α + h _y (α) approximates, whereby k _y (α) and h _y (α) are variable depending on α, so that the difference is formed from the estimated values of the input variables given by the filters of each filter pair, whereby two functions g ( h _yv ; h _yh ) or h (h _yv ; h _yh ) can be obtained from the equation system with two unknowns h _yv and h _yh and also at least one size of the following sizes k _yv , k _yh and ω, v _{y are taken} as an estimate from the last-mentioned filter (E-AkF) and that, if appropriate, the other movement variables (such as the slip angle β, the slip angle α _v , α _h and the _Lateral forces F _yv , F _yh ) can be reconstructed.