DE19957187B4 - Method and device for crash detection - Google Patents

Abstract

Method for crash detection, in which at least one sensor signal (S1-S6; S) generated by a sensor (2a-2f) of a sensor unit (2) is fed to an evaluation device (4), in which the evaluation device (4) derives from the sensor signal supplied to it (S1-S6; S) determines a crash-characteristic pattern and generates a control signal (S _T ) correlated to the determined pattern, which initiates the activation of at least one safety device (S _E ), characterized in that the at least one of the evaluation device (4) A weighting signal (W1-W6; W) is ascertained by summing the amplitude of the sensor signal (S1-S6; S) and the amplitude of at least one time derivative of the sensor signal (S1-S6; S) is carried out, and that from the weighting signal (W1-W6) thus formed, the crash-characterizing pattern is derived, and that the evaluation device (4) generates the crash character thus generated Rising pattern with patterns stored in the evaluation device (4) compares and, with a match within predetermined deviations of the determined pattern with one of the stored patterns, the control signal ...

Description

The The invention relates to a method for crash detection, in which at least a sensor signal generated by a sensor of a sensor unit fed to an evaluation is, in which the evaluation device from the supplied sensor signal a crash characteristic pattern determined and one to be determined Pattern correlated control signal generates the activation initiated at least one safety device.

One such method and one suitable for carrying out this method Device are known. They serve, in case of an accident of the Motor vehicle detection of the accident, especially in terms the course of the accident and the severity of the accident, and / or an accident classification, in particular the type of intrusion of the colliding with the motor vehicle Object, determine and appropriate safeguards for the Initiate occupants of the motor vehicle. In the execution of the procedure it is important that on the one hand a sufficiently fast and reliable Chrashdetektion takes place, and that on the other hand a non-mandatory triggering of the safeguards due the associated occupant hazard is prevented.

The DE 3803426 A1 discloses a method for activating a safety system, which proposes to calculate from the signal of the accelerometer the occupant advancement occurring during the crash.

The DE 3924507 A1 discloses a method for triggering restraint means in which an acceleration signal is measured and a speed signal is derived by temporal integration of the acceleration signal. To form a triggering criterion for the restraint means, a threshold value for the speed signal is set, which is variable.

It is therefore an object of the present invention, a method and a device of the type mentioned in such a way, that in simpler Way a fast accident detection and / or accident classification is achievable.

These The object is achieved by the method according to the invention that of the Evaluation device from the at least one sensor signal supplied to it a weighting signal is determined by a summation of the Amplitude of the sensor signal and the amplitude of at least one temporal Derivation of the sensor signal is performed, and that from the thus formed weighting signal the crash characterizing pattern is derived, and that the Evaluation the thus generated, crash-characterizing Compare patterns with patterns stored in the evaluation device and at a match within predetermined deviations of the detected pattern with one of the stored patterns, the control signal for the Security device generated, wherein as crash sensors of a sensor unit Piezo-type crash sensors are used, each one of its mechanical Load proportional electrical sensor signal, so that at a deformation of a vehicle component on which the sensor unit is arranged, the sensor signal supplied by the sensor unit This deformation follows in its course, so that the occurred deformation by the utilization of the direct piezoelectric effect immediately into a measurable one implemented electrical signal size becomes.

By the measures according to the invention will created in an advantageous manner, a method which characterized by a fast crash detection and a simple crash classification. Another advantage of the method according to the invention consists in its easy to implement, because for crash detection and classification only easy to perform temporal Derivatives of the sensor signal generated by the sensor unit for Determination of a weighting signal for determining the crash characterizing Pattern as well as a comparison of this pattern with stored patterns required is.

A advantageous development of the invention provides that the crashcharakterisierendes Pattern by a weighted summation of the amplitude of the sensor unit generated sensor signal and at least one time derivative of this signal is formed, preferably it is provided that in the weighted accumulation the sensor signal as well as the first and enter the second time derivative of the sensor signal. A Such accumulation of these individual contributions has the advantage that it is through the appropriate choice of weighting factors of individual posts easily possible is to simply consider certain crash scenarios.

A Another advantageous embodiment of the invention provides that only each a section of the identified pattern falling within a certain time window included in the comparison. Such a measure has the advantage that thereby in simple way a temporal resolution of the accident process is made possible.

A further advantageous embodiment of The invention provides that non-overlapping time windows are used. Such a measure has the advantage that in this way accident events represented by the pattern and / or the sub-pattern (s) are resolved into a chronological sequence of discrete events.

A Another advantageous embodiment of the invention provides that a sliding Time window is used. Such a measure has the advantage that thereby simply a correlation of the individual periods of the accident easily feasible is.

Further advantageous developments of the invention are the subject of the dependent claims.

Further Details and advantages of the invention are the embodiment to be taken, which is described below with reference to the figures. Show it:

1 an embodiment of a device for crash detection together with a motor vehicle,

2 a functional diagram of the device,

3 a digrammatic representation of six sensor signals in a first accident,

4 a schematic representation of the six sensor signals in a second accident,

5 a schematic representation of one of the sensor signals together with derived signals and

6 a schematic representation of a crash pattern.

The method for crash detection is described below with reference to the in 1 illustrated embodiment of a corresponding device and at the in 2 illustrated functional diagram explained. The 1 shows a vehicle F, which in the side region of the passenger compartment G each with a sensor unit 2 is equipped, wherein the preferably identically constructed sensor units 2 six crash sensors each 2a - 2f exhibit. It will be apparent to those skilled in the art that the arrangement of six crash sensors described herein 2a - 2f per sensor unit 2 has only exemplary character. It is also possible to have more or less of these crash sensors 2a - 2f provided. Furthermore, the applicability of the method described and the device developed thereafter is not limited to a corresponding processing system for detecting a side impact. It is Z. B. also possible that a sensor unit 2 is arranged in the front and / or in the rear area. As crash sensors 2a - 2f piezoelectric crash sensors are preferably used in this case, each of which generates a mechanical load proportional to their electrical sensor signal S1-S6, so that in a deformation of the vehicle component on which the sensor unit 2 is arranged, that of the sensor unit 2 supplied sensor signal S of this deformation follows in its course, so that the occurred deformation is converted by the utilization of the direct piezoelectric effect directly into a measurable electrical signal magnitude.

The vehicle F has except the sensor units 2 Furthermore, a well-known and therefore no longer shown precrash sensor 5 and a safing sensor system 6 on which for the plausibility check of the sensor units 2 generated sensor signals S is used. In the case shown here, the safing sensor system 6 two in itself known lateral acceleration sensors 6a . 6b on, the sensor signal S _S to the evaluation 4 is directed.

In 2 is now the interaction of these individual components of the device 1 shown. The ones from the crash sensors 2a - 2f each sensor unit 2 generated individual sensor signals S1-S6 are via a preferably designed as a sensor bus signal line 3 to an evaluation device 4 passed through the then one in 1 only schematically shown safety devices S _{E of} the vehicle F, z. B. side airbags, activating control signal S _{T can be} generated.

Only schematically illustrated sensors 5 ' the precrash sensor system 5 , in particular radar sensors, serve to determine the impact time point and / or the impact zone and / or the relative velocity of the object colliding with the vehicle F. The of the precrash sensor 5 generated sensor signals P _S are the evaluation 4 , in particular an airbag control unit, the device 1 Among other things, they serve to determine the time zero of the subsequent signal processing.

The case of a collision from the sensors 2a - 2f generated sensor signals S1-S6, which together the sensor signal S of the sensor unit 2 are training, now in the 3 and 4 shown, in 3 typical sensor signals S1-S6 are shown as they occur when a foreign vehicle collides laterally at a speed of 50 km / h with the vehicle F, so if a breitflä chiger impact takes place. In 4 are the corresponding signals S1-S6 of the crash sensors 2a - 2f the sensor unit 2 illustrated in the case that a pile at a speed of about 30 km / h in the passenger compartment G side of the motor vehicle F occurs, so that a singular collision takes place. Here, in both figures, the solid line represents the sensor signal S1 of the first crash sensor 2a , Correspondingly, the lines S2 (dashed line), S3 (punctured), S4 (dot-dashed), S5 (dashed double) and S6 (double-dashed dot) characterize the corresponding sensor signals S2-S6 of the crash sensors 2 B - 2f ,

That of the sensor unit 2 generated from the sensor signals S1-S6 of the sensors 2a - 2f existing and preferably digitized sensor signal S of the sensor unit 2 will now - as already described - the evaluation device 4 for further processing: For each of these individual signals S1-S6 is now by the evaluation 4 , a differential or differential time derivative is formed, as can be seen from the 5 is apparent. This in 5 upper diagram 5a shows the amplitude u (t) one of the evaluation 4 supplied signals S1-S6, this signal having been digitized at a sampling frequency, which is typically in the kHz range. In the following it is assumed in an exemplary manner that this is the case of the first crash sensor 2a generated sensor signal S1 acts. This does not limit the generality of the following effects, since the further sensor signals S2-S6 of the sensor unit 2 in a corresponding manner as the first sensor signal S1 are processed. The evaluation device 4 determined from the first crash sensor 2a generated sensor signal S1, a time derivative of the sensor signal S1 representing signal S1 ', whose amplitude d _ν (t) in diagram 5b is shown and determined according to the formula d _ν (t) = S1 (t) - S1 (t - Δt). In a corresponding manner, the signal S1 '' is derived from the signal S1 'representing the speed of the deformation by a further derivation of time, whose amplitude d _νν (t) In 5c is shown. The signal S1 "formed in accordance with d _νν = S1 (t) -S1 (t-2Δt) represents the temporal change of the deformation speed or acceleration.

In order to be able to generate a crash-characterizing pattern from these three aforementioned signals S1, S1 ', S1 ", it is provided that these three signals are superposed, preferably weighted, with the resulting weighting function having an amplitude W (t) = | a * S1 (t) | + | b · d _ν (t) | + | c · d _νν (t) | in 5d is shown. In the case shown here, a = c = 1/2 and b = 1 were chosen. These weighting constants a, b, c are in this case preferably determined empirically or by model calculations or by a discriminant analysis.

the It is clear to a person skilled in the art that the inclusion of only the first sensor signal S1 and its two derivatives S1 ', S1' 'in the weighting signal W1 only exemplary Character possesses. So it is quite possible that for certain applications only the signals S1, S1 'or also at least the third time derivative of the signal S1 or a defined selection from the group of the original one Signal S1 'and at least one time derivative in the determination of the weighting signal W1 arrives.

In the diagram 6a of the 6 The weighting signal W1 for the sensor signal S1 determined as described above and the correspondingly generated weighting signals W2-W6 for the further sensor signals S2-S6 are shown. In order to be able to derive a crash-characterizing pattern from these weighting signals W1-W6, which each have an amplitude w (t), it is provided here that, for a given number of time intervals Δt, it is examined for each weighting signal W1-W6 whether its amplitude w (t ) is above or below a predefined threshold S _w . In the case of an amplitude w (t) which lies below the first threshold value S _w in a specific time interval, a first, low value is assigned to a pattern element of the crash-characterizing pattern representing this time interval, and a second, high value is assigned to an amplitude lying above the threshold value S _w , By way of example, it is possible to set the first, lower value "0" and the second, higher value of this only two-stage distinction with "1". Accordingly, the pattern elements of the first sensor signal S1 are in the pattern thus generated for the first two time intervals "0", in the next five time intervals "1", and in the subsequent time interval again "0". In a corresponding manner, the weighting signal W2 becomes the sequence of the pattern elements "0", "0", "0", "1", "1", "1", "1" and "1". Accordingly, the pattern elements derived from the weighting signal W3 associated with the third sensor signal S3 are five times "0" and then three times "1", those of the fourth weighting signal W4 are six times "0", then "1", and then again in time "0", those of the fifth weighting signal are "0" six times and then "1" twice. The partial pattern obtained from the weighting signal W6 accordingly has eight pattern elements with the value "0".

In the above description, it has been assumed that only a two-stage evaluation of the weighting signals W1-W6 takes place here, such that all the amplitudes w (t) below the first threshold value S _{w have} the low value (here: 0) for each time interval above this first threshold value S _w amplitudes to the higher value (here: 1) ordered. It is also possible to carry out a more than two-stage signal evaluation, such that a weighting signal W, which lies above the first threshold value S _w but below a higher, second threshold value, continues to be assigned the abovementioned second value (here: 1) but a signal whose amplitude is above the second threshold in a time interval no longer receives the aforementioned value but the higher third value (eg: 2). In this way, slightly n-level values can be used to refine the crash pattern.

The thus generated crash pattern is now with the in the evaluation 4 stored crash patterns compared. This can be done by elemental comparison of the corresponding crash patterns. But it is also possible that only a portion of the corresponding crash pattern is compared with each other, in such a way that in a first step, the values of the first pattern elements are compared with each other, that in the second step, the subsequent values of the next n-pattern Elements are compared with each other, etc.

It but it is also possible that about the crash pattern an m-crash pattern-wide time window is being moved, and always only the respective pattern elements lying in the time window in the Compare.

Now results in the comparison of the thus generated crash pattern with that in the evaluation 4 stored crash patterns a match within predefined limits match, interpreted by the evaluation 4 this as a sure indication that the accident pattern underlying the stored pattern is present. She then initiates the appropriate safeguards.

Claims (16)

Crash detection method in which at least one of a sensor ( 2a - 2f ) a sensor unit ( 2 ) generated sensor signal (S1-S6; S) of an evaluation device ( 4 ), in which the evaluation device ( 4 ) determines a crash-characteristic pattern from the sensor signal supplied to it (S1-S6; S) and generates a control signal (S _T ) correlated to the determined pattern, which initiates the activation of at least one safety device (S _E ), characterized in that the evaluation device ( 4 A weighting signal (W1-W6; W) is ascertained from the at least one sensor signal fed thereto (S1-S6; S) by summing the amplitude of the sensor signal (S1-S6; S) and the amplitude of at least one time derivative of the sensor signal (S1-S6; S) is carried out, and that the crash-characterizing pattern is derived from the weighting signal (W1-W6) thus formed, and in that the evaluation device ( 4 ) the thus generated crash-characterizing pattern with in the evaluation ( 4 ) and, in the case of a conformity of the detected pattern with one of the stored patterns, within a given deviation, generates the control signal (S _T ) for the safety device (S _E ), and that as a sensor ( 2a - 2f a piezo-crash sensor is used which generates an electrical sensor signal (S1-S6; S) which is proportional to its mechanical load, so that during a deformation of a vehicle component on which the sensor unit ( 2 ) arranged by the sensor unit ( 2 ) follows this deformation in its course, so that the occurred deformation is converted by the utilization of the direct piezoelectric effect directly into a measurable electrical signal magnitude. Method according to Claim 1, characterized in that the crash-characteristic pattern is derived in such a way that it can be read by the evaluation device for a number of predefined time intervals ( 4 ) examines whether the amplitude of the weighting signal (W1-W6) is above a predefined threshold value (S _w ). Method according to one of the preceding claims, characterized characterized in that Weighting signal (W1-W6) by summing the amplitude of the Sensor signal (S1-S6) and the amplitude of the first time derivative the sensor signal (S1-S6) is formed. Method according to claim 3, characterized that in the summation for determining the weighting signal (W1-W6) further the amplitude of the second time derivative of the sensor signal (S1-S6) is included. Method according to one of the preceding claims, characterized characterized in that in the summation for the determination of the weighting signal (W1-W6) incoming amplitudes of the sensor signal (S1-S6) and the at least a time derivative of the sensor signal (S1-S6) of each of these posts weighted with a weighting factor. Method according to one of the preceding claims, characterized in that the evaluation device ( 4 ), it is examined whether the amplitude of the weighting signal (W1-W6) is above the predefined threshold value (S _w ) and below a further threshold value, or above this higher threshold value, compared to the first-mentioned threshold value (S _w ). Method according to one of the preceding claims, characterized in that a sensor unit ( 2 ) with several sensors ( 2a - 2f ) is used, that in each case one of these sensors ( 2a - 2f ) of the sensor unit ( 2 ) generates a sensor signal (S1-S6), and that a weighting signal (W1-W6) is generated for each of these sensor signals (S1-S6). Method according to one of the preceding claims, characterized characterized in that Crash-characterizing patterns from the individual sensor signals (S1-S6) obtained weighting signals (W1-W6) is composed. Method according to one of the preceding claims, characterized characterized in that crashcharakterisierende pattern is formed like a matrix, wherein the elements of a row or column of this pattern matrix the time sequence of the obtained from a weighting signal (W1-W6) Pattern elements is formed. Method according to one of the preceding claims, characterized in that the comparison of the determined pattern with one of the in the evaluation device ( 4 ) also stored in matrix form patterns by a column and row by row comparison. Method according to one of the preceding claims, characterized characterized in that Pattern comparison is done such that in each case to m time intervals associated Pattern elements the pattern matrix are compared. Method according to one of the preceding claims, characterized characterized in that Pattern comparison each falling within a specific time window Pattern elements of the two matrix-like processed pattern with each other be compared. Method according to one of the preceding claims, characterized characterized in that a sliding time window is used. Method according to one of the preceding claims, characterized characterized in that a non-overlapping Time window is used. Crash detection device comprising an evaluation device ( 4 ) comprising at least one of a sensor ( 2a - 2f ) a sensor unit ( 2 ) can be supplied (S1-S6; S), wherein the evaluation device ( 4 ) determines a crash-characteristic pattern from the sensor signal supplied to it (S1-S6; S) and generates a control signal (S _T ) correlated to the determined pattern, which initiates the activation of the at least one safety device (S _E ), characterized in that the evaluation device ( 4 ) determines a weighting signal (W1-W6; W) from the at least one sensor signal supplied thereto (S1-S6; S) by summing the amplitude of the sensor signal (S1-S6; S) and the amplitude of at least one time derivative of the sensor signal (S1). S1-S6; S) is carried out, and in that the evaluation device ( 4 ) the thus generated crash-characterizing pattern with in the evaluation ( 4 ) and, when the conformity of the detected pattern with one of the stored patterns is within predetermined deviations, generates the control signal (S _T ) for the safety device (S _E ), and that 2a - 2f a piezo-crash sensor is used which generates an electrical sensor signal (S1-S6; S) which is proportional to its mechanical load, so that during a deformation of a vehicle component on which the sensor unit ( 2 ) arranged by the sensor unit ( 2 ) follows this deformation in its course, so that the occurred deformation is converted by the utilization of the direct piezoelectric effect directly into a measurable electrical signal magnitude. Vehicle characterized by a device according to the preceding claim.