DE10003739A1 - System parameters identification in vehicle, involves obtaining movement equations with respect to measured speed and acceleration based on vehicle substitute model - Google Patents

Abstract

Speed and acceleration of the vehicle are measured and the vector of these values is determined. Movement equations with respect to the measurement values are obtained based on a vehicle substitute model. The ratio of estimated and measurement values is weighted by optimization parameter which is determined based on the covariance matrix.

Description

The invention relates to a method and a system for identi System parameters in vehicles after the upper beg Reef of claim 1 and 10 respectively.

To assess the driving behavior of vehicles, übli Usually in driving mode or on a test bench measured from the vehicle and from the measurements obtained Characteristic values determined by means of the driving behavior of the vehicle can be characterized, for example the self-steering gradient, the slip angle gradient or the roll angle gradient. Based these parameters can determine the driving behavior of the vehicle Time range or frequency range can be assessed. To the ge To be able to gain the desired parameters, a ma thematic vehicle replacement model can be formulated with the equations of motion approximate the driving behavior of the vehicle is to be described approximately. The vehicle replacement model is usually constructed in such a way that at least the cross vehicle dynamics - lateral acceleration and yaw acceleration supply - are to be recorded. Taking into account the measurands and the equations of motion of the vehicle replacement model the parameters sought from the parameters of the replacement model be determined.

The parameters of interest must be accurate enough be determined so that a reliable statement about the driving behavior of the vehicle based on these parameters is. To determine the parameters with the required quality  To be able to, with the usual identification mo dents to determine system parameters during the measurement drives a wide range of driving maneuvers and under various road stimuli are covered to ensure safe make that the identification model for the parameters identification required data can be supplied. The variety of driving maneuvers under different conditions conditions are time-consuming and not always necessary reproducible accuracy.

Identification procedure based on a linear single-track model of a vehicle are, for example, in the print Writings DE 42 26 749 A1 and DE 43 25 413 A1 described wor the.

The invention is based on the problem of identification system parameters of a vehicle to improve or simplify. One possibility in particular is to be given the measuring effort without impairing the quality of the Reduce parameters.

This problem is solved according to the invention with the features of the Proverbs 1 and 10 solved.

According to the new procedure for the identification of system pa parameters in vehicles uses a parameter estimation method, which is expedient on the method of covariance intersection which is a further development of the Kalman filter for Estimation, filtering and data fusion applications poses. However, the covariance intersection method is in the Un Unlike the Kalman filter, it was able to identify a parameter to carry out high-quality fication even if Estimated values of the process and error or noise components cor relate. The covariance intersection allows the fusion of Quantities whose degree of correlation is unknown.  

According to the invention it is provided that the System parameters an iterative calculation rule in which an optimization parameter is taken into account with which the shares from the estimate and from the measurement weighted differently depending on the quality of the measured values can be tet. The optimization parameter is over an optimization function according to a given functional certainly.

The equations of motion of the vehicle replacement model are ins particular about the calculation rule of covariance Intersection method considered. Usually it is enough Single-track vehicle replacement model for determining the lateral dynamics and the roll dynamics of a vehicle for determining the System parameters.

The method based on the covariance intersection method for the identification of system parameters offers the advantage that the Be. to be met in each iteration step the optimization parameter is one of the quality of the current one Weighting corresponding to measurement values in the iteration algorithm between estimates and measurements. Thereby does the estimation method work even with a low or a lack of excitation without worsening the results for the system parameters, since in the case of a low or wrong lent suggestion of the estimated proportion in the calculation rule weighted significantly more for parameter identification is called the measurement portion. Conversely, when the measurement contains much more information than the estimate, accordingly, the measurement portion is weighted more than that Estimate.

Another advantage of using the covariance intersection The process is that the process results diverge  in contrast to previously known methods in Re gel can be excluded.

In particular, the self-steering gradient, the slip angle gradient and the roll angle gradient are determined. The measurands to be recorded during the measurement trip are in particular the longitudinal speed of the vehicle, the transverse acceleration inclination, the yaw rate and, if necessary, the roll speed, the steering angle and the cross speed speed. These measurands are used to determine the system parameters ter used.

The system parameters identified online can be used for evaluation of driving behavior can be used. Beyond that it is also possible after the parameter identification Fahr maneuvers of the vehicle replacement model in time or frequency richly simulate in order to make another or more accurate reviewer direction of driving behavior.

The system according to the invention for performing the method holds a measuring device with which the required ones are in the vehicle Vehicle state variables can be measured or determined, and a computing unit, which is also arranged in the vehicle and in which the measurement recorded in the measuring device sizes in particular using the covariance Intersection method can be evaluated. In this version are the system parameters that characterize the driving behavior are available online in the vehicle and can be used according to a be preferred training for the generation of control signals he be drawn about the driving behavior of the vehicle can be influenced. Such control signals are used for example the actuators of an active, controllable driving plant, an anti-lock braking system, a drive slip Control system, an engine control or a transmission control supply. This enables you to  To be able to positively manipulate driving behavior in the event that the parameters characterizing the driving behavior outside a defined area, which means constructive deviation can also be compensated for by an ideal value such as production or assembly errors or wear and tear while driving stuff.

Further advantages and practical embodiments are the further claims, the description of the figures and the drawing refer to the flow chart with the process step for measuring state variables, identifying systems parameters and generation of driving behavior influencing Control signals shows.

The covariance intersection method, which is used for the online identification of system parameters θ in vehicles, is based on an iterative calculation rule based on the equations

P ^-1 _{k + 1} = ωP ^-1 _k + (1 - ω) H ^T _{k + 1} R ^-1 H _{k + 1}

P ^-1 _{k + 1} θ _{k + 1} = ωP ^-1 _k θ _k + (1 - ω) H ^T _{k + 1} R ^-1 y _{k + 1} .

Here, "k" denotes the current iteration step index, "P" a covariance matrix, "θ" to be determined telenden system parameters, "R" a generally constant Mo dell variance matrix, "H" one of measured values and the motion equations of the underlying vehicle replacement model pending Jacobi matrix, "ω" an optimization parameter and "y" Measured values, where "k" and "ω" are scalar quantities, "y" and "θ" Denote vectors and are "P", "H" and "R" matrices.

The covariance intersection procedure is carried out iteratively online in the vehicle. For this purpose, the vehicle is equipped with a measuring device for measuring vehicle state variables and a computing unit for evaluating the measurement results and, if appropriate, for generating control signals which are to be supplied to vehicle assemblies which influence driving behavior. The measurement includes the following vehicle state variables, which are summarized in a measurement vector z:

z _{k + 1} = [y ^T _{k + 1} , u ^T _{k + 1} ] ^T ,

where "y" denotes the measured values of the system outputs and "u" denotes the measured values of the system inputs. The measured values y of the system outputs correspond to the equations of motion h of the mathematical vehicle replacement model according to the context

y = h (u _{k + 1} , θ).

The measured values u of the system inputs flow directly into the Calculate the right side of the equations of motion.

In the case of a single-track vehicle model taking into account the vehicle lateral dynamics and the vehicle roll dynamics, the vector of the measured values is y of the system outputs

y = [a _y , d ² Ψ / dt ² , ϕ] ^T

with the lateral acceleration a _y , the yaw acceleration d ² Ψ / dt ² and the roll angle ϕ. The equations of motion h are calculated in the computing unit according to the stored substitute model; the equations of motion h depend on the system parameters θ and the measured values u of the system inputs, with u in the single-track model in the measurement vector

u = [v _x , v _y , dΨ / dt, dϕ / dt, δ _H ] ^T

the longitudinal vehicle speed v _x , the transverse vehicle speed v _y , the yaw rate dΨ / dt, the roll speed dϕ / dt and the steering angle δ _H must be taken into account.

The covariance matrix P and the system parameters θ in the covariance intersection algorithm are determined iteratively based on the starting values P ₀ , θ ₀ that are to be specified. The Jacobian matrix H is according to the context

H ^T _{k + 1} = ie ^T (u _{k + 1} , θ) / dθ

as a differential or as a quotient of differences from the be equations h and the system parameters θ are determined.

The optimization parameter ω, which determines the ratio of estimated values to measured values in the covariance intersection method, is expediently determined in such a way that the optimization criterion

det (P _{k + 1} ) = minimum,

according to which the determinant of the covariance matrix P is a minimum should be fulfilled. In alternative versions, a Be also indicated, different cost functions for determining the optimization parameter ω.

In summary, the covariance intersection method can be calculated using the equation theorem, taking into account appropriate transformations for a single-track vehicle model

θ _{k + 1} = θ _k + K _{k + 1} [y _{k + 1} - h (u _{k + 1} , θ _k )]

K _{k + 1} = (1 - ω) P _k H ^T _{k + 1} [H _{k + 1} P _k H ^T _{k + 1} + R] ^-1

H ^T _{k + 1} = ie ^T (u _{k + 1} , θ) / dθ

P ^-1 _{k + 1} = ωP ^-1 _k + (1 - ω) H ^T _{k + 1} R ^-1 H _{k + 1}

ω off: det (P _{k + 1} ) = minimum
describe.

For clarification, the algorithm for the first two iteration steps is listed with k = 0 and k = 1:
k = 0 (initialization step):

θ ₁ = θ ₀ + K ₁ [y ₁ - h (u ₁ , θ ₀ )]

K ₁ = (1 - ω) P ₀ H ^T ₁ [H ₁ P ₀ H ^T ₁ + R] ^-1

H ^T ₁ = dh ^T (u ₁ , θ) / dθ

P ^-1 ₁ = ωP ^-1 ₀ + (1 - ω) H ^T ₁ R ^-1 H ₁

ω off: det (P ₁ ) = minimum,
where the model variance matrix R has constant values and θ ₀ , P ₀ are specified as starting values for the system parameters or the covariance matrix. With "K" is a correction matrix.
k = 1:

θ ₂ = θ ₁ + K ₂ [y ₂ - h (u ₂ , θ ₁ )]

K ₂ = (1 - ω) P ₁ H ^T ₂ [H ₂ P ₁ H ^T ₂ + R] ^-1

H ^T ₂ = dh ^T (u ₂ , θ) / dθ

P ^-1 ₂ = ωP ^-1 ₁ + (1 - ω) H ^T ₂ R ^-1 H ₂

ω off: det (P ₂ ) = minimum.

The flow chart shown in the figure illustrates the How the covariance intersection method works Functioning of the system integrated in a vehicle, in which this method for online identification of vehicle pa rametern is implemented.

In a first process step 1, the index k representing the iteration step is initially initialized to the value zero at the beginning of the method. In the next step 2, measured values are recorded via the measuring device integrated in the vehicle when the vehicle is moving, in particular state variables of the vehicle at the position, speed and acceleration level. The vector of the measured values z _{k + 1 is} determined, which probably includes system output measured values y _{k + 1} as well as system input measured values u _{k + 1} . Following the measurement, the equations of motion h are determined in sequence step 3, taking into account the measured values y _{k + 1} and u _{k + 1} , which are based on the vehicle replacement model currently in use. The optimization parameter ω is also calculated, in particular from the condition that the determinant of the covariance matrix takes a minimum. According to the covariance intersection method, the system parameters θ _{k + 1 are} determined, which expediently comprise at least the self-steering gradient, the slip angle gradient and the roll angle gradient.

In step 4, the system parameters determined are subjected to a quality check. If the absolute value | θ _{k + 1} | of a system parameter lies outside a permissible, predetermined range θ _limit , the "yes" branching continues in accordance with sequence step 6 and an actuating signal S _{St is} generated in the computing unit in order to manipulate an actuator of a vehicle unit with which the driving dynamics Properties of the vehicle can be influenced. For this, actively controllable components of the chassis come into question, but also components via which the condition of the brakes, the engine or the transmission can be influenced. Manipulating an aggregate influencing the driving condition can compensate for deviations in the system parameters from an ideal value online in the vehicle.

After the control signal S _St has been generated, the process proceeds to step 5, in which the index k is increased by the value one. The method is then run through again, starting with step 2, the measurement of the vehicle state variables.

If the absolute value | θ _{k + 1} | a system parameter in the query for sequence step 4 is within the permissible, predetermined range θ _limit , the "no" branch is accordingly moved to sequence step 5 to increase the index k by the value one and finally to the beginning of the method returned to step 2.

It may be appropriate to use the method according to the Sequence step 3, the measurement of the state variables and the determ development of the system parameters, cancel and the Er knowledge of the system parameters of a constructive adjustment solution of the vehicle.

Claims (16)

1.Procedure for identifying system parameters in vehicles, measured values (y, u) representing vehicle state variables are measured while the vehicle is in motion and evaluated in a computing unit in accordance with a calculation rule for determining the system parameters (θ), with equations of motion in the calculation rule (h) of a vehicle calculation model and both estimated values and measured values (y, u) are taken into account, characterized in that the ratio of estimated values and measured values in the calculation rule for parameter identification is weighted via an optimization parameter (ω), the weighting in the Way that

• - in the event that the measurement in the current measurement step contains more information than the estimate, the measurement is taken into account more and
• - In the event that the measurement does not provide new information in relation to the estimate, the estimate is taken into account more.

2. The method according to claim 1, characterized in that
that the optimization parameter (ω) is determined in such a way
that a function of the covariance matrix (P) corresponds to a given functional:
f (P _{k + 1} ) = functional. 3. The method according to claim 2, characterized in that
that the optimization parameter (ω) is determined in such a way
that the determinant of the covariance matrix (P) has a minimum:
det (P _{k + 1} ) = minimum. 4. The method according to any one of claims 1 to 3, characterized in
that the calculation rule according to the context
P ^-1 _{k + 1} = ωP ^-1 _k + (1 - ω) H ^T _{k + 1} R ^-1 H _{k + 1}
P ^-1 _{k + 1} θ _{k + 1} = ωP ^-1 _k θ _k + (1 - ω) H ^T _{k + 1} R ^-1 y _{k + 1}
is carried out iteratively, in which
k the index showing the current iteration step
P is a covariance matrix
θ the system parameters to be determined
R is a model variance matrix
H is a Jacobian matrix dependent on measured values
ω the optimization parameter and
y the measured values
referred to and the Jacobian matrix (H) as a function of measured values (u) and equations of motion (h) from the differential or the difference quotient
H ^T _{k + 1} = ie ^T (u _{k + 1} , θ) / dθ
is determined, the optimization parameter (ω) being determined by means of the optimization function. 5. The method according to any one of claims 1 to 4, characterized in that in the equations of motion (h) at least the transverse acceleration (a _y ) and the yaw acceleration (d ² Ψ / dt ² ) are taken into account. 6. The method according to any one of claims 1 to 5, characterized in that the roll acceleration (d ² ϕ / dt ² ) is taken into account in the equations of motion (h). 7. The method according to any one of claims 1 to 6, characterized in that one or more of the following variables are measured: the longitudinal speed (v _x ), the lateral speed (v _y ), the lateral acceleration (a _y ), the yaw rate (dΨ / dt), the roll speed (dϕ / dt) and the steering wheel angle (δ _H ). 8. The method according to any one of claims 1 to 7, characterized, that the model variance matrix (R) in the calculation rule to determine the system parameters (θ) is constant. 9. The method according to any one of claims 1 to 8, characterized,  that the procedure is carried out online in the vehicle. 10. System for performing the method according to one of the An Proverbs 1 to 9, with a measuring device and an arithmetic unit unit in the vehicle in which it is recorded in the measuring device measured values to determine the driving behavior system parameters (θ) can be evaluated. 11. System according to claim 10, characterized in
that steep signals (S _St ) can be generated in the computing unit to influence the driving behavior of the vehicle,
that the determined system parameters (θ) are outside a defined range. 12. System according to claim 11, characterized in that the control signals (S _St ) actuators of an active chassis can be supplied. 13. System according to claim 11 or 12, characterized in that the control signals (S _St ) are an anti-lock system feed bar. 14. System according to any one of claims 11 to 13, characterized in that the control signals (S _St ) can be fed to a traction control system. 15. System according to any one of claims 11 to 14, characterized in that the control signals (S _St ) can be fed to an engine control. 16. System according to any one of claims 11 to 15, characterized in that the control signals (S _St ) can be fed to a transmission control.