DE102006009372A1 - Method for analysing and evaluating signal, particularly sensor signal, involves carrying out frequency analysis of unfiltered signal for determining frequency spectrum - Google Patents

Abstract

The method involves carrying out a frequency analysis of an unfiltered signal (S2) for determining a frequency spectrum. A frequeny portion in the determined frequency spectrum is evaluated (S3) for determining a close spectral signal power. The determined close spectral signal power is set (S4) relative to a filtered signal. Independent claims are also included for the following: (1) a device for analysing and evaluating signal (2) a controller for a passenger protective system of a vehicle.

Description

The The invention relates to a method for analyzing and evaluating a Signal, in particular a sensor signal for detecting a rollover process a vehicle according to claim 1 and a corresponding device according to claim 14.

For the detection of a rollover process of a vehicle, the signals of a rotation rate sensor are generally evaluated around the vehicle longitudinal axis (x-axis) as well as a lateral (y) and a vertical (z) acceleration sensor. From the EP 0 942 854 B1 For example, it is known to generate a trigger signal for airbags by a logical AND operation of the yaw rate and the lateral acceleration. This should allow a very early and safe release of occupant protection.

Some Protective means such as side airbags, such as a head airbag or a so-called airbag curtain ("curtain") in a rollover process triggered in time be that the airbags between the vehicle and the due the vehicle movement in sideways shifting inmates Completely can unfold. For example have to Head airbags are fired so early, that they are between the head of an occupant and the vehicle roof, the spars and the B-pillar of the vehicle can unfold. Become The head airbags may not unfold in time and completely, may under circumstances the retention effect severely impaired be or maybe by the action of the retention means an additional Risk of injury occur.

The lateral impact of the vehicle on a curb or similar Obstacle) or the lateral, sliding friction of the vehicle on a subsoil ("soil-trip") are operations in which a timely ignition of airbags is particularly critical. In these cases, it is very difficult Use sensor signals to determine whether the vehicle is overturning or not, as usually at the time of ignition or triggering of a Airbags the vehicle is just starting to move around its longitudinal axis, i.e. to rotate the x-axis. Dependent From the vehicle speed, the vehicle can overturn or not. By the mere Evaluation of the rate of rotation is not possible to determine for sure whether a rollover process will be done or not. It has also been shown that inclusion the lateral acceleration in the generation of a trigger decision not always the required improvement in terms of early and safe triggering brings, especially the lateral acceleration sensor signals as the rotation rate sensor signals usually heavily low-pass filtered be to detect long-term movements of the vehicle. Indeed Low pass filtering means a loss of information for a subsequent one Analysis and evaluation of a signal, in particular with regard to a triggering decision a protection means of an occupant protection system of a vehicle.

Out International patent application WO 2005/100101 A2 are a method and an apparatus for analyzing and evaluating a signal, in particular a sensor signal known, especially for use in a control unit for a Occupant protection system for influencing triggering criteria and thresholds suitable.

task The present invention is now an improved process and an apparatus for analyzing and evaluating a signal, in particular a sensor signal for detecting a rollover process to propose a vehicle.

These The object is achieved by a method for analyzing and evaluating a Signal, in particular a sensor signal for detecting a rollover process a vehicle having the features of claim 1 and a corresponding one Device solved with the features of claim 14. preferred Embodiments of the invention will become apparent from the dependent claims.

An essential idea of the invention is to analyze the spectral power of an unfiltered signal and to evaluate the filtered signal on the basis of the analyzed spectral power. The invention preferably proves useful in analyzing and evaluating signals from acceleration sensors for measuring lateral and / or vertical axis acceleration of a vehicle in critical rollover events, at an appropriate time and time for vehicle occupant protection system protection means such as side or head airbags to determine for sure. When evaluating filtered acceleration sensor signals, a safe and early triggering decision can often not be made at all, especially in the case of the critical rollover processes mentioned in particular, since the high spectral components required for this have already been filtered out of the signals. The fact that in the invention in the evaluation of the filtered sensor signal, an analysis of high-frequency sensor signal components flows, which contain much more information than the filtered sensor signal, it is possible to make an early and relatively safe triggering decision. In particular, the invention improves a time-correct, ie a separation of triggering events and non-triggering events corresponding to the triggering or ignition time of protective means, in particular when only at the required triggering time Low rotation rates are present and based on these low rotation rates a safe separation of triggering event and non-triggering event is hardly possible.

• a) eine Frequenzanalyse des ungefilterten Signals zum Ermitteln des Frequenzspektrums durchgeführt wird,
• b) die Frequenzanteile im ermittelten Frequenzspektrum zum Ermitteln einer angenäherten spektralen Signalleistung bewertet werden, und
• c) die ermittelte angenäherte spektrale Signalleistung in Relation zu einem gefilterten Signal gesetzt wird.

The invention now according to one embodiment relates to a method for analyzing and evaluating a signal, in particular a sensor signal, wherein

• a) a frequency analysis of the unfiltered signal for determining the frequency spectrum is performed,
• b) the frequency components in the determined frequency spectrum are evaluated to determine an approximated spectral signal power, and
• c) the determined approximate spectral signal power is set in relation to a filtered signal.

As already mentioned above, The invention is preferably suitable for the analysis and evaluation of Signals from acceleration sensors of occupant protection systems for vehicles, especially when it comes to early Detection of a rollover process of a vehicle, for example due to a curb or soil trip.

According to one embodiment The invention can for frequency analysis, the unfiltered signal means an orthogonal decomposition into basic functions of low order or equivalent Method to be transformed. This also as Orthogonal-DeComposition (ODC) known method is in the already mentioned international patent application WO 2005/100101 A2 of the applicant described in detail. It owns across from Another method of frequency analysis has the advantage that it has a less computational effort than, for example, a discrete Fourier or a wavelet transformation needed and therefore no special signal processor for implementation needed, but also executed with microprocessors can be.

According to one another embodiment of the invention assessed the frequency components in the determined frequency spectrum in terms of amount and about window integrals be summed, resulting in an approximate spectral signal power can be determined. For example, in the case of analysis the signals from acceleration sensors in the transverse or vertical direction a vehicle thereby the approximate spectral signal power determined in transverse or vertical direction and after another Processing into a triggering decision for protective equipment, how head or side airbags flow.

According to one embodiment The invention can provide the approximated spectral signal power be compared to the filtered signal in that it is compared with at least one frequency threshold which depends on the filtered one Signal is set. The comparison with a frequency threshold can be implemented with little technical effort, for example in hardware by means of a comparator or in software by one simple comparison command.

According to one embodiment the frequency threshold can be designed dynamically. Especially it can be raised when the filtered signal exceeds a threshold, and reduced when the filtered signal falls below the threshold. This can improve the accuracy of the analysis and evaluation of the signal be improved as long-term trends as movements of a vehicle be recognized by evaluation of the filtered signal and in the Analysis and evaluation of the unfiltered signal by the variation to enter the threshold can.

In an embodiment The invention can be easily processed as a result the analysis and evaluation of the signal a logical evaluation result signal be activated for further evaluation, if the spectral signal power the Frequency threshold exceeds. This evaluation result signal may be, for example, from a control unit of a Occupant protection system of a vehicle as another trigger decision-relevant Criterion, for example a triggering threshold for a Signal of a rotation rate sensor to vary. If the assessment result signal is activated, for example, the control unit, the triggering threshold for the rotation rate sensor signal reduce, so even at a lower rotational speed of the vehicle by e.g. its longitudinal axis Protection means are activated. This allows a side and head airbag already early triggered be to keep vehicle occupants safe from injury due to high Protect lateral accelerations in the initial stage of a rollover process.

According to one another embodiment of the invention the unfiltered signal and the filtered signal from different Signal sources come. For example, frequency ratings of lateral accelerations or lateral acceleration sensor signals compared or in relation to frequency ratings of vertical accelerations or vertical acceleration sensor signals are set, in particular to generate a decision to modify thresholds for trigger sizes.

According to a further embodiment, the invention relates to a method for detecting a rollover process of a vehicle, in which an unfiltered signal from a sensor for measuring a Acceleration of the vehicle in the lateral or vertical direction according to a method according to the invention and analyzed and evaluated as described above and the evaluation result is used for the detection of the rollover process. As already explained above, the method according to the invention for analyzing and evaluating a signal is particularly suitable for the particularly early detection of a rollover event of a vehicle and can therefore be advantageously used for reliable and reliable rollover detection.

• a) Überprüfen, ob eine gemessene Drehrate des Fahrzeugs größer als eine erste Schwelle ist,
• b) Überprüfen, ob das Bewertungsergebnis einen Überrollvorgang signalisiert, falls die Drehrate größer ist als die erste Schwelle, und
• c) Überprüfen, ob die Drehrate größer als eine zweite Schwelle ist, falls das Bewertungsergebnis einen Überrollvorgang signalisiert.

According to one embodiment of the invention, the method can be provided, for example, for implementing an additional triggering path for the rollover detection and for this purpose have the following steps:

• a) checking whether a measured rate of turn of the vehicle is greater than a first threshold,
• b) checking whether the evaluation result signals a rollover event, if the rotation rate is greater than the first threshold, and
• c) Check if the yaw rate is greater than a second threshold, if the evaluation result signals a rollover event.

This additional triggering path can be next to a conventional one Triggering path, which is checked only once, whether the turning rate of the vehicle is greater than the first threshold is to be implemented and so a premature one release protection of the occupant protection system than with the conventional one triggering path possible cause.

• a) Überprüfen, ob das Bewertungsergebnis einen Überrollvorgang signalisiert,
• b) Modifizieren einer Schwelle für eine gemessene Drehrate des Fahrzeugs, und
• c) Überprüfen, ob die Drehrate größer als die modifizierte Schwelle ist, falls das Bewertungsergebnis einen Überrollvorgang signalisiert.

According to an alternative embodiment of the invention, the method may also be provided for modifying a triggering threshold and for this purpose comprise the following steps:

• a) checking whether the evaluation result signals a rollover event,
• b) modifying a threshold for a measured rate of turn of the vehicle, and
• c) Check if the yaw rate is greater than the modified threshold, if the evaluation result signals a rollover event.

hereby becomes a conventional one triggering path with a one-time review of one threshold crossing improved by the yaw rate by the (trigger) threshold depending on Evaluation result is modified, for example, lowered or is reduced if the evaluation result is a rollover process signaled.

According to one another embodiment The invention may be dependent from the evaluation result also a protective agent triggering threshold for a filtered signal from a sensor to measure acceleration the vehicle in lateral or vertical direction or a rotation rate sensor for detecting the rollover process be set.

According to one embodiment The invention may provide the unfiltered signal from a sensor for Measuring an acceleration of the vehicle in the lateral direction and the filtered signal from a sensor for measuring acceleration of the vehicle in the vertical direction or vice versa, i. in other words, that the filtered signal of an acceleration sensor with the frequency weighting of an unfiltered signal from another Acceleration sensor can be evaluated, causing the triggering of Protective means can be better adapted to a driving situation.

Especially can according to a embodiment the invention, the triggering threshold be reduced if, as a result of the analysis and evaluation of the unfiltered signal, the logical evaluation signal is activated. This means that due to the analysis and evaluation of the unfiltered Signals a trigger decision-relevant Event was detected, resulting in a reduction in the trigger level leads, at the time of a trip of a protective agent in advance.

According to a further embodiment, the invention relates to a device for analyzing and evaluating a signal, in particular a sensor signal, with
Analysis means, which are designed to perform a frequency analysis of the unfiltered signal for determining the frequency spectrum,
first evaluation means, which are designed to evaluate the frequency components in the determined frequency spectrum for determining an approximated spectral signal power, and
second weighting means configured to set the detected approximate spectral signal power in relation to a filtered signal.

In an embodiment can the analysis means, the first and second evaluation means a programmatically designed according to microprocessor be implemented. As already explained above, especially in a Frequency analysis by means of the ODC method the computing power of a commercial Microprocessors to be fully adequate for implementation.

Especially can according to a embodiment the invention of the microprocessor programmatically set up further or programmed to a method according to the invention and how perform as described above to be able to.

Finally, the invention according to another embodiment relates to a control device for an in Sassenschutzsystem a vehicle having a device according to the invention and as described above.

Further Advantages and applications The present invention will become apparent from the following description in conjunction with the embodiments illustrated in the drawings.

In the description, in the claims, in the abstract and in the drawings are those in the listed below List of reference symbols used terms and associated reference numerals used.

The Drawings show in:

1 a flowchart of an embodiment of a method for analyzing and evaluating an acceleration sensor signal according to the invention;

2 a block diagram of an occupant protection system for rollover detection, comprising an embodiment of an apparatus for analyzing and evaluating the signals of a transverse and a Hochachsenbeschleunigungssensors according to the invention;

3 a flowchart of a first embodiment of a method for roll over detection of a vehicle based on a rotation rate signal, wherein an embodiment of a method for analyzing and evaluating an acceleration sensor signal according to the invention used as an additional trigger path that is more sensitive than the rotation rate signal based trip path;

4 a flowchart of a second embodiment of a method for roll over detection of a vehicle based on a rotation rate signal, wherein an embodiment of a method for analyzing and evaluating an acceleration sensor signal according to the invention is used to modify the rotation rate triggering threshold;

5 a timing diagram with exemplary courses of determined approximate spectral signal powers for a trigger and a non-triggering event;

6 FIG. 5 is a timing diagram illustrating exemplary traces of filtered acceleration sensor signals for a trigger and a non-trigger event; FIG. and

7 An embodiment of an evaluation diagram based on a filtered acceleration sensor signal as an evaluation size and based on an unfiltered acceleration sensor signal analyzed and evaluated according to the invention as a further evaluation variable.

1 shows the flowchart of an algorithm for implementing the inventive method in an occupant protection system of a vehicle. This algorithm can be carried out by a commercially available microprocessor or correspondingly powerful microcontroller, as used for example in control units for occupant protection systems. For this purpose, the algorithm is stored in the form of executable program code, for example in a non-volatile memory such as a ROM (Read Only Memory) or a PROM (Programmable ROM). Instead of implementing the algorithm in software, it can also be implemented in special hardware, for example as an ASIC (Application Specific Integrated Circuit) or in the form of a PGA (Programmable Gate Array).

Of the Algorithm processes an unfiltered acceleration sensor signal, For example, a signal of an acceleration sensor for detecting a Acceleration of a vehicle in the direction of the vehicle transverse axis or -Hochachse, to finally a time-correct separation of tripping and non-triggering events to support, which, above all, as already explained at the beginning in so-called Curb and soil trip operations the vehicle alone based on an evaluation of a yaw rate sensor signal or a filtered acceleration sensor signal is hardly possible.

First, will in step S1, the unfiltered acceleration sensor signal with a Digitized analog-to-digital converter to digitally process it to be able to. The digitization should as possible with a sampling rate, in which also high-frequency signal components be digitized in the acceleration sensor signal, since these signal components for the following Analysis and evaluation of the signal are of particular importance. In particular, the digitized acceleration sensor signals should the high-frequency signal components that contain a filtered Acceleration sensor signal usually does not contain (such as explained in the beginning).

In a subsequent step S2, the digitized unfiltered acceleration sensor signal for frequency analysis is orthogonally decomposed into basic functions of low order (ODC method). Details of the ODC method and the use of the ODC method for occupant protection systems can be found in the already mentioned international patent application WO 2005/100101 A2 of the applicant. After performing the step S2 is a frequency spectrum of the unfiltered digitized acceleration sensor signal.

in the Step S3 will now be the frequency components of the acceleration sensor signal assessed in terms of magnitude in the determined frequency spectrum, and the frequency components will be over then Window integrals integrated to an approximate spectral signal power to investigate. The evaluation and integration is also done by the ODC procedure.

Now will approximate the thus determined spectral signal power in relation to the unfiltered acceleration sensor signal set to obtain a criterion for a triggering decision of a protective agent of the occupant protection system is relevant. The in-relation setting takes place in that the determined approximated spectral signal power is compared with at least one frequency threshold (step S4). This depends the frequency threshold of the filtered acceleration sensor signal and becomes dependent set by this signal.

exceeds the approximate Spectral signal power, the frequency threshold (step S5), is a logical evaluation result signal activated as the criterion for the triggering decision a protective agent can be used. Will the frequency threshold not exceeded, the logical evaluation result signal remains deactivated.

As the logical evaluation result signal in the final triggering decision protection of the occupant protection system in the following with reference to further embodiments of the invention explained.

At first, however, shows 2 a part of an occupant protection system of a vehicle intended for roll over detection. For this purpose, the occupant protection system has a rotation rate sensor 18 on, which detects the rate of rotation of the vehicle about the vehicle longitudinal or x-axis. Furthermore, a lateral acceleration sensor 12 for detecting a transverse axis or y-acceleration of the vehicle and a vertical acceleration sensor 14 provided for detecting a Hochachsen- or z acceleration. That of the sensor 12 generated lateral acceleration signal and that of the sensor 14 generated vertical acceleration signal are each with a low pass 22 respectively. 24 filtered and by means of an analog-to-digital converter 32 respectively. 38 digitized. In parallel, the unfiltered lateral and vertical acceleration signals are each by means of an analog-to-digital converter 34 respectively. 36 digitized.

The digitized unfiltered acceleration signals and the digitized filtered acceleration signals are stored in a microprocessor 20 further processed, the analysis means 26 which are adapted to perform a frequency analysis of the unfiltered acceleration signals for determining the corresponding frequency spectra, first evaluation means 28 which are adapted to evaluate the frequency components in the determined frequency spectra for determining corresponding approximate spectral signal powers of the unfiltered acceleration signals, and second evaluation means 30 implemented to set the determined approximate spectral signal powers in relation to the corresponding filtered acceleration signals. The means 26 . 28 and 30 can be implemented in the form of software routines, for example, the steps of a method as described in 1 shown and described above implement. Alternatively, the funds can 26 . 28 and 30 also be implemented in hardware. In this case, the reference number would 20 denote a corresponding digital signal processing circuit instead of a microprocessor.

The second evaluation means 30 give a judgment result signal as a result of the analysis and evaluation of the acceleration sensor signals 40 from a control unit 16 the occupant protection system, on the input side, the rotation rate sensor signal and the filtered acceleration sensor signals are further processed. In particular, the evaluation result signal is used 40 in addition, a triggering threshold for the rotation rate sensor signal in the control unit 16 to modify. In the simplest case, the evaluation result signal 40 a binary signal, for example, at a logic high level the controller 16 indicates that the triggering threshold should be reduced because a curb or soil trip has been detected due to the analysis and evaluation of the unfiltered acceleration sensor signals. In the modification of the triggering threshold for the rotation rate sensor signal are also the filtered acceleration sensor signals, as will be explained in more detail below. Finally, depending on the supplied signals, the control unit generates a triggering signal 42 for a protection means, for example a head or side airbag.

The Elements 20 . 26 . 28 and 30 can also be in the control unit 16 be integrated or by a correspondingly powerful microprocessor or controller of the controller 16 be implemented.

3 FIG. 10 shows a flowchart of a method in which the analysis and evaluation of an unfiltered acceleration sensor signal is used as an additional triggering path in a vehicle occupant protection system to be more sensitive to the detection of rollovers of the vehicle. First, it is checked in a step S11 whether a rotation rate of the vehicle to the sen longitudinal axis or a corresponding rotation rate sensor signal exceeds a first omega threshold 1 (omega stands for the rotation rate). If the rate of rotation exceeds this threshold, in a further step S14 the plausibility of the rotation rate sensor signal is checked by a further independent sensor, for example by the signal of a second rotation rate sensor. If this check also reveals that the threshold has been exceeded, in a step S15, a triggering request for a rollover protection means such as, for example, an automatic roll bar, belt tensioners, side and head airbags and the like for protection against rollover, is provided. If it is determined in step S11 that the rotation rate is still below the first omega threshold 1, a further query is performed in a step S12. It checks whether the frequency criterion is fulfilled. In this case, frequency criterion is understood to be the result of the analysis and evaluation of the unfiltered acceleration sensor signal according to the invention. In particular, the frequency criterion can correspond to the assessment result signal described above, ie it is then checked whether the assessment result signal is active or inactive. If the frequency criterion is not fulfilled or the evaluation result signal is inactive, the algorithm is terminated. Otherwise, another query follows in a step S13, in which it is checked whether the rotation rate exceeds a second omega threshold 2. The second omega threshold 2 is in this case selected to be smaller than the first omega threshold 1. Thus, an exceeding of this smaller omega threshold 2 is only checked by the rotation rate if the frequency criterion is met, ie, the unfiltered acceleration sensor signal can be deduced to indicate a rollover event , If this is the case, the procedure continues with the already explained step S14, otherwise the algorithm is terminated.

4 FIG. 12 shows a further flow chart of a method in which the analysis and evaluation of an unfiltered acceleration sensor signal is used to modify a triggering threshold for a measured rate of rotation of the vehicle about its longitudinal axis. As a result, the rotation rate triggering threshold or omega threshold can be better adapted to a driving situation of the vehicle, for example if the result of the analysis and evaluation of the unfiltered acceleration sensor signal is that the vehicle is in a curb or soil trip operation. In the flowchart of the illustrated algorithm, the frequency criterion is first checked in a step S21 (see the description of this 3 ). If the analysis and evaluation of the unfiltered acceleration sensor signal indicates that the frequency criterion has been met, a modification of the omega threshold takes place in a step S23, for example a reduction or reduction of the threshold, so that tripping of a threshold is already effected at lower rotation rates than without consideration of the frequency criterion Protective agent can be done. In a following step S22, the rotation rate is then checked for exceeding the possibly modified omega threshold. If the threshold is exceeded by the rotation rate, the plausibility of the rotation rate sensor signal is checked by a further independent sensor in a step S24, for example by the signal of a second rotation rate sensor. If this check also reveals that the threshold has been exceeded, a triggering request for a roll-over protection means such as an automatic roll bar, belt tensioners, side and head airbags and the like for roll-over protection is provided in a step S25.

5 shows the time course 100 an approximate spectral signal power of an acceleration sensor signal for a trigger event (solid bold line) and the time course 102 an approximate spectral signal power of an acceleration event signal for a non-triggering event (dashed line), in which although high values of the approximated spectral signal power can occur as shown, but does not cause injury to occupants or a rollover process. Likewise, in the diagram, the time course 104 a frequency threshold that is variable. This frequency threshold can be used to determine if a trigger-relevant event has occurred. The diagram in 5 is related to the in 6 read diagram. In the diagram of 6 are the ones in the diagram of 5 shown courses corresponding time courses 106 respectively. 108 of the filtered acceleration sensor signal is also shown for the triggering event (solid bold line) and the non-triggering event (dashed line). Likewise is the time course 110 an acceleration threshold shown. Based on the gradients 106 and 108 It can be seen that an exceeding of the acceleration threshold occurs almost at the same time. According to this, a distinction between the triggering and non-triggering events would be virtually impossible on the basis of these courses alone. However, exceeding the acceleration threshold can be used to set the frequency threshold in the graph of 5 by increasing the frequency threshold so that, as in the diagram of 5 shown, trigger and non-trigger event based on the gradients 100 and 102 can be distinguished. In this variant of the method according to the invention, therefore, the frequency criterion is a criterion in the "frequency vs. Acceleration diagram, "as explained in the self-explanatory 7 is shown.

Of course they are Further applications of the invention in particular in connection with Occupant protection systems possible and thinkable. The invention allows the more accurate, in particular more timely control of protection means of an occupant protection system and can work with conventional components, such as For example, a correspondingly powerful microprocessor implemented become.

10

Occupant protection system of a vehicle

12

Lateral acceleration sensor

14

Vertical acceleration sensor

16

Control device for protective equipment of the occupant protection system

18

Yaw rate sensor

20

microprocessor

22 24

Low Pass Filter

26

Frequency analysis means

28

first evaluating means

30

second evaluating means

32 34, 36, 38

Analog to digital converter

40

Rating result signal

42

trigger signal

S1-S6

steps

S11-S15

steps

S21-S25

steps

100 102

time courses approximate spectral signal powers

104

time Course of the frequency threshold

106 108

time courses of acceleration sensor signals

110

time Course of an acceleration threshold

Claims (17)

Method for analyzing and evaluating a Signal, in particular a sensor signal, wherein a) a frequency analysis the unfiltered signal for determining the frequency spectrum is performed (S2), b) the frequency components in the determined frequency spectrum to determine an approximate spectral signal power are evaluated (S3), and c) the determined approximate spectral signal power in relation to a filtered signal is set (S4). Method according to claim 1, characterized in that that for frequency analysis, the unfiltered signal by means of an orthogonal Decomposition into base functions of low order or equivalent Method is transformed (S2). Method according to claim 2, characterized in that that for determining an approximated spectral Signal power the frequency components in the determined frequency spectrum assessed in terms of value and via window integrals be summed (S3). Method according to claim 1, 2 or 3, characterized that the approximated spectral Signal power to the filtered signal thereby set in relation will be compared to at least one frequency threshold that becomes dependent is adjusted by the filtered signal (S4). Method according to claim 4, characterized in that a frequency threshold ( 104 ) is raised when the filtered signal ( 106 . 108 ) a threshold ( 110 ) and decreases when the filtered signal ( 106 . 108 ) the threshold ( 110 ) falls below. Method according to claim 4 or 5, characterized in that a logical evaluation result signal ( 40 ) is activated as a result of the analysis and evaluation of the signal for further evaluation (S6) when the spectral signal power exceeds the frequency threshold (S5). Method according to one of the preceding claims, characterized characterized in that the unfiltered signal and the filtered Signal come from different signal sources. Method for detecting a roll-over operation of a vehicle, in which an unfiltered signal from a sensor for measuring a Acceleration of the vehicle in lateral or vertical direction analyzed and evaluated according to a method of the preceding claims becomes (S1-S6) and the evaluation result for the detection of the rollover process is used (S12; S21). Method according to claim 8, marked by the following steps: a) Check if a measured rate of turn of the vehicle greater than a first threshold is (S11), b) Check if the evaluation result a rollover process signals (S12) if the yaw rate is greater than the first threshold, and c) Check if the spin rate is greater than a second threshold is (S13) if the evaluation result is a rollover event signaled. A method according to claim 8, characterized by the following steps: a) checking whether the evaluation result signals a rollover event (S21), b) modifying a threshold for a measured rate of turn of the vehicle (S23), and c) Checking that the yaw rate is greater than the modified threshold (S22) if the evaluation result signals a rollover event. Method according to claim 8, characterized in that that dependent from the evaluation result, a protection trigger threshold for a filtered signal a sensor for measuring an acceleration of the vehicle in lateral or vertical direction or a rotation rate sensor for detection of the rollover process is set. Method according to claim 11, characterized in that that the unfiltered signal from a sensor to measure an acceleration of the vehicle in the lateral direction and the filtered signal from a sensor for measuring an acceleration of the vehicle in vertical Direction or vice versa. Method according to claim 11 or 12, characterized that reduces the triggering threshold becomes (S23) when as a result of the analysis and evaluation of the unfiltered Signal the logical evaluation signal is activated (S21). Contraption ( 20 ) for analyzing and evaluating a signal, in particular a sensor signal, with analysis means ( 26 ), which are designed to perform a frequency analysis of the unfiltered signal for determining the frequency spectrum, first evaluation means ( 28 ), which are designed to evaluate the frequency components in the determined frequency spectrum for determining an approximated spectral signal power, and second evaluation means ( 30 ) configured to set the detected approximate spectral signal power in relation to a filtered signal. Apparatus according to claim 14, characterized in that the analysis means ( 26 ), the first and second evaluation means ( 28 . 30 ) by a program technically equipped microprocessor ( 20 ) are implemented. Device according to claim 15, characterized in that the microprocessor ( 20 ) is also technically set up or programmed in order to be able to carry out a method according to one of Claims 1 to 13. Control unit ( 16 ) for an occupant protection system ( 10 ) of a vehicle, characterized in that it comprises a device ( 20 ) according to one of claims 14 to 16.