DE102007040912A1 - Sensor-signal characteristic processing method for motor vehicle, involves eliminating interference portion from signal characteristic, and processing usage portion remained after elimination in chassis frame-regulation system - Google Patents

Abstract

The method involves eliminating an interference portion occurring with a temporal periodicity from a signal characteristic, while a non-periodic signal-portion is maintained with a frequency consistence, where the signal characteristic is detected by a yaw rate sensor or a transverse oscillation sensor. A phase correction and an amplitude correction with respect to the interference portion are carried out. A usage portion of the signal characteristic remained after the elimination is further processed in a chassis frame-regulation system.

Description

The The invention relates to a method for processing a means a sensor detected temporal waveform, in particular a driving dynamic characteristic of a Motor vehicle reproducing signal waveform for further use in one Chassis control system.

generally known modern motor vehicles are equipped with chassis control systems, which the longitudinal dynamics and / or the lateral dynamics and / or the vertical dynamics of the vehicle can influence. at the longitudinal dynamic Control systems may, for example, be a brake control system, in a lateral dynamic control system of a preferably two-lane Motor vehicle, for example, a in the steering system of the vehicle intervening control system and in a vertical dynamic control system eg. one such, which is the damper rates and / or spring rates in the so-called. Suspension can influence. With a regulatory intervention in the steering system of the vehicle For example, the steering assistance torque provided to the driver of the vehicle may be provided be set or adjusted. The latter can, for example dependent on from the yaw rate or lateral acceleration (known to those skilled in the art) of the vehicle under consideration be regulated by the driving speed, these only exemplified sizes (yaw rate, Lateral acceleration driving speed) under the selected generic term a "dynamic driving Characteristic "fall and means suitable and suitably installed in the vehicle sensors become.

uniformities in particular on a wheel (or wheel tire) of the vehicle may be periodic Vibrations in the temporal signal course of a measured dynamic driving Cause size. Such a nonuniformity may, for example, in terms of mass distribution or the stiffness distribution or just with regard to the geometry of, for example, a wheel; with the latter is the possible one so-called imbalance generally known. From such irregularities resulting periodic oscillations in the temporal signal curve a driving dynamic size can by Resonances partly a big one reinforcement experienced and thus the actual measurement of this dynamic driving Size considerably distort, what a negative impact on a suspension control system, which These signals can be used, and in particular its Can reduce robustness. To have the necessary robustness over the To ensure the life of the vehicle often becomes a limitation the efficiency of the suspension control system accepted.

In this regard one To show improvement is the object of the present invention. It is expressly indicated that a proposed and described here Process for the preparation of a determined by means of a sensor time waveform on the application in suspension control systems limited is, although based on this application, the explanation of the present invention. Rather, an inventive method is everywhere applicable, where a temporal by means of a sensor detected waveform next to a so-called useful portion of interest, a periodic, for the Further processing of the sensor signal uninteresting so-called interference component contains. This can only be mentioned as an example when measuring in storage of prime movers or their output torque be the case; usually always where a periodic movement takes place, from which a periodic interference component (for example in the form of Imbalance) can be contained in some sort of sensor signal can.

The solution the task mentioned is for a method according to the preamble of claim 1, characterized that at least one occurring with a temporal periodicity so-called noise component from the sensor waveform is eliminated while a non-periodic signal component with a noise component comparable Frequency composition is retained before the after this elimination remaining so-called useful portion of the sensor signal waveform further processed becomes. According to a preferred algorithm that does this, will in a temporally narrow analysis time window of no periodicity Useful portion of the original Sensor waveform as an approximation considered linear over time and by high-pass filtering greatly reduced. From the remaining remaining part of the original Sensor waveform is caused by the high-pass filtering Phase shift and amplitude modification determined, whereupon a Phase correction and an amplitude correction with respect to this disturbance component carried out and then the correspondingly corrected course of the noise component from the original one Sensor waveform is subtracted to the useful portion of the original sensor waveform receive.

It is thus proposed to remove at least one significant exclusively periodic component from a time signal waveform, for example a driving dynamic parameter. Unlike the use of conventional bandpass filters thus non-periodic components remain with partially the same Frequency composition as the periodic components obtained virtually unchanged in the waveform. In this way - in contrast to a conventional bandpass filtering - sudden transient changes, for example, a lateral acceleration signal of the vehicle undergoes no phase shift and amplitude modification and can thus be considered genuine in a chassis control system, while superimposed on the waveform periodic vibration, the for example resulting from a tire imbalance, at least greatly reduced.

at single application of the above-proposed (and explained in more detail below) preferred algorithm is, if multiple periodic noise components with different periodicities contained in the sensor waveform, the one with the largest amplitude eliminated. One or more additional interference component (s) with a lower amplitude can or can thus be eliminated one after the other by this proposed Algorithm is applied several times in succession.

In In a preferred application, said analysis time window in the order of 0.5 seconds, especially if the useful portion of the original Sensor waveform in a chassis control system, which determines the steering torque in the steering system of the vehicle, further processed or taken into account and the sensor waveform from a yaw rate sensor or a lateral acceleration sensor of the Vehicle originates.

Attached is a Signal flow diagram in which a preferred method according to the invention is shown. It is based on an analysis time window of 0.5 seconds, for example, a caused by a steering action of the driver Transverse acceleration or yaw rate within this analysis time window approximately considered as linear. The temporal waveform (the lateral acceleration sensor or yaw rate sensor) superimposed is a harmonic distortion caused by tire imbalance, for example, where resonance influences a clear reinforcement this suggestion, usually in the order of magnitude between 5 Hz and 30 Hz.

Thus, approximately for the temporal waveform, the following applies x 1 (t) = a 0 + a 1 t + A · sin (ω D · T), equation 0 where ω _D denotes the angular frequency and A denotes the amplitude of a harmonic periodic disturbance in the form of a so-called interference component [A · sin (ω _D · t)]. The so-called useful portion, ie the signal progression of the lateral acceleration or yaw rate actually caused by a steering action of the driver, is described by the term [a ₀ + a ₁ t].

We are now looking for the following time course of the noise component x G (t) = A · sin (ω D · T), equation 1 so that finally by the following subtraction y (t) = x 1 (t) - x G (T) the periodic interference component of the signal curve is eliminated and only the following so-called useful component remains: y (t) = a 0 + a 1 t.

The time signal waveform x ₁ (t) containing the useful component and the interference component is thus passed through a first high-pass filter in which the useful component is essentially removed, so that only the interference component x ₂ (t) remains. which is passed through another high-pass filter in order to be able to determine the phase shift and amplitude modification caused by the first high-pass filter, as explained below; After that the course x ₃ (t) results.

ermittelt. Mit Kenntnis derer kann im nächsten, mit „Variable Transport Delay" bezeichneten Block eine Zeitverzögerung und somit eine Phasenkorrektur des Signalverlaufs x₂(t) durchgeführt werden, woraus sich der Verlauf

ergibt. Dabei kann der mit „Variable Transport Delay" bezeichnete Block ein aus Matlab/Simulink von der Fa. Mathworks^TM bekannter Systemblock sein oder durch die Funktion x(k) = Z^–k beschrieben sein, mit

und Ta als Abtastzeit.In the subsequent block, a possibly existing periodicity is determined by cross-correlation of x ₂ (t) and x ₃ (t) - a significant periodicity is known for pronounced minima and maxima in the cross-correlation - and their angular frequency ω _D - this is it is that of the periodic noise component - as well as the time shift caused by said first high pass filter

determined. Knowing this can be done in the next block called "Variable Transport Delay" a time delay and thus a phase correction of the waveform x ₂ (t) are performed, resulting in the course

results. In this case, the block designated by "Variable Transport Delay" may be a system block known from Matlab / Simulink from Mathworks ^TM or may be described by the function x (k) = Z ^-k , with

and Ta as sampling time.

durchgeführt wird. Danach kann dieser zeitliche Signalverlauf vom ursprünglichen Signalverlauf x₁(t) subtrahiert werden, um den reinen Nutzanteil hiervon (weiter oben mit y(t) bezeichnet) zu erhalten.Parallel to this, the amplitude modification G _HP (ω _D ) resulting from the first high-pass filtering is preferably determined via a table (so-called "lookup table high-pass filter") with knowledge of the excitation frequency ω _D , from which an amplitude correction with respect to the signal profile is obtained by multiplication by its reciprocal value x ₂ , more precisely with regard to the already phase-corrected course

is carried out. Thereafter, this time waveform can be subtracted from the original waveform x ₁ (t) to obtain the net payload thereof (denoted by y (t) above).

It should be expressly understood that this method of course works even with non-harmonic periodic noise component, although in the following a mathematical proof for the operation of the proposed method is performed only for a harmonic interference component:
To determine the interference component x _G (t) from the temporal signal curve x ₁ (t), first the linear component a ₀ + a ₁ t from x ₁ (t) by high-pass filtering x ₂ (t) = x ₁ (t) · g _HP (t) is greatly reduced, where g _HP (t) represents the weighting function of the high-pass filter and "*" the convolution is symbolized by the time function The high-pass filter is, for example, a second-order filter with a corner frequency of 5 Hz for the treated use case, so that the following applies: (a ₀ + a ₁ t) · g _HP (t) ≈ 0.

It follows from Equation 2: x ₂ (t) ≈ A _HP · sin (ω _D · t + φ) with the phase advance φ due to the high-pass filter and modified amplitude A _HP .

Analog follows x 3 (t) = x 2 (T) · g HP (T) x 3 (t) ≈ A HPHP * Sin (ω D · T + 2φ).

The transfer function the high-pass filter is known. In the Laplace range, the transfer function is for example.

With s = jω, the amplitude modification as well as the phase shift of a harmonic signal can be determined analytically by the high-pass filter at a given angular frequency ω _D :

To reduce the computation time, it is recommended to use a so-called table ("LOOK-Up Table"), which provides the amplitude modification at the given excitation frequency ω = ω _D :

By cross-correlation of x ₂ (t) and x ₃ (t), the excitation frequency of the periodic disturbance and the phase shift caused by the high-pass filter can be determined:

For easier calculation, the cross-correlation is shown over an infinite time range. The resulting error is sufficiently small.

mit k ∊ N.A search algorithm determines the maximum next to the zero shift.

with k ε N.

With k = 1, τ _{min describes} the abscissa of the minimum next to the maximum with the abscissa τ _max .

folgt die Frequenz der periodischen Störung:

From the difference of the shift

follows the frequency of the periodic disturbance:

With ω _D , | G _HP (ω _D ) | determined by lookup table and divided by x ₂ . By a delay of x ₂ by the time τ _max follows for

From Equation 6, Equation 3 and Equation 4 give Equation 7:

And thus: x G (t) ≈ Asin (ω D · T)

From this it can be seen that the presented algorithm or the method presented here causes a strong reduction of the periodic components in the measurement x ₁ (t), wherein it should be pointed out that quite a number of details can be deviated from the above explanations, without departing from the content of the claims.

Claims (4)

A method for processing a signal waveform determined by means of a sensor, in particular a signal characteristic reproducing a driving dynamic parameter of a motor vehicle for further use in a chassis control system, characterized in that at least one so-called interference component occurring with a temporal periodicity is eliminated from the sensor signal waveform , while a non-periodic signal component is retained with a frequency composition comparable to the interference component, before the so-called useful component of the sensor signal profile remaining after this elimination is further processed. Method according to claim 1, characterized in that that in a temporally narrow analysis time window of no periodicity having Useful portion of the original Sensor waveform as approximately temporally viewed linearly and greatly reduced by high-pass filtering and that of the remaining residual part of the original Sensor waveform caused by the high-pass filtering Phase shift and amplitude modification is determined, whereupon a phase correction and an amplitude correction with respect to this disturbance component carried out and then the correspondingly corrected course of the noise component from the original one Sensor waveform is subtracted to the useful portion of the original sensor waveform to obtain. Method according to claim 1 or 2, characterized that these procedures are carried out several times in succession, periodic disturbances, the different periodicities have to eliminate from the sensor waveform. Method according to one of the preceding claims, characterized by at least one of the following characteristics: • the original one (Sensor) waveform comes from a yaw rate sensor or a Lateral acceleration sensor • of the Useful portion of the original Sensor waveform is used in a suspension control system adjusts the steering torque in the steering system of the vehicle, further processed or considered • the analysis window is on the order of magnitude of 0.5 seconds