EP2387516A1

EP2387516A1 - Method and controller for controlling personal safety devices for a vehicle

Info

Publication number: EP2387516A1
Application number: EP09768003A
Authority: EP
Inventors: Jens Becker; Marcus Hiemer
Original assignee: Robert Bosch GmbH
Current assignee: Robert Bosch GmbH
Priority date: 2009-01-13
Filing date: 2009-11-23
Publication date: 2011-11-23
Also published as: WO2010081580A1; US8660757B2; DE102009000160B4; CN102271969B; US20120004811A1; CN102271969A; DE102009000160A1

Abstract

A method and a controller for controlling personal safety devices for a vehicle are disclosed, wherein at least one sensor signal from an accident detector is produced by an interface. At least two features are generated from the sensor signal, presented on a single value range. Control of the personal safety devices is carried out on the basis of the at east two presented features.

Description

description

title

Method and control device for controlling personal protective equipment for a vehicle

State of the art

The invention relates to a method and a control device for controlling personal protection means for a vehicle according to the preamble of the independent claims.

From DE 10 2007 004 345 Al a method for the control of personal protection means is known in which at least one feature is extracted from at least one size and the activation decision takes place as a function of a crash classification, wherein the

Crash classification depending on the at least one feature takes place. The activation of the personal protection means takes place in dependence on this activation decision. The drive decision is formed by the fact that a sequence control is provided which, depending on at least one course variable, has a plurality of functions for the

Crash classification activates or deactivates and / or determines which at least one characteristic is used for the respective function. The historical variable can be, for example, the time from the beginning of the crash or the feature or an event.

Disclosure of the invention

The inventive method and the control device according to the invention for controlling personal protection means for a vehicle with the features of the independent claims have the advantage that Now the at least two features that were formed from the at least one sensor signal, are mapped to a single range of values. Depending on at least these two mapped features, the drive decision is formed. This single range of values for the at least two features allows the individual features to be weighted equally, allowing for better classification of a crash event, especially when more than two features, ie, three-dimensional or higher-dimensional viewing, are involved. Ie. the same weighting influences the evaluation of these characteristics mapped to the single value range. In particular, this improves the classification in so-called machine learning based methods such as neural networks, fuzzy logic, vector quantization, maximum likelihood and support vector machines, as well as other such machine learning based methods. For machine-learning based methods, reference is made to a publication by Hastie et al .: T. Hastie, R. Tibshirani, J. Friedman;

The Elements of Statistical Learning - Data Mining, Inference and Prediction, Springer Verlag Berlin, 2001

Furthermore, this single range of values for the features used allows a flexible exchange of features without the range of values of these individual

To consider features. For example, if a feature does not sufficiently contribute to algorithm performance, such an exchange may improve functionality.

The particular advantage with equally weighted features for learning-based

The method is because such learning-based methods use the statistical properties of a signal. One of these properties is, for example, the signal variance. If the range of values of one feature is significantly larger than that of another, these statistical properties are distorted. The best results are obtained by a corresponding normalization, for example to a value range of -1 to +1. It is also advantageous to use a range of 0 to 1, since the algorithm features used here are predominantly positive and thus a bit in the range of values can be saved compared to the range of -1 to +1, so that a better resolution can be used. In particular, the present method also optimized for computing resources such as runtime, RAM, ROM and especially EEPROM consumption be designed. And this despite a high accuracy.

Furthermore, the method according to the invention or the control unit according to the invention makes it possible to freely select the accuracy.

In the present case, the activation of personal protection means, for example the energization of ignition elements of airbags, which lead to their ignition and thus to the inflation of such an airbag, is understood by the activation of personal protection means. Therefore one understands also under

Personal protective passive passenger protection such as airbags, belt tensioners, roll bars, headrests, but also active personal protection such as a vehicle dynamics control system or a brake system.

Providing the at least one sensor signal of an accident sensor system is understood herein to mean the provision of the at least one sensor signal.

The at least one sensor signal of an accident sensor system may be analog or digital signals, a multiplex of signals. In the

Accident sensors include an acceleration sensor, an air pressure sensor, an environment sensor such as a radar, video and / or ultrasonic sensors and a structure-borne noise sensor as examples in question.

Producing the at least two features from the at least one

Sensor signal may be understood to be a filtering of mathematical operations such as derivative, integration or multiple integration or multiple derivative or other arithmetic operations. An example of such a feature is, for example, the integrated acceleration of an acceleration sensor. The two features are generally different, but they can be identical.

By mapping the at least two features to a single range of values, one must understand the normalization of these two features to this single range of values. Ie. it will be a value range from 0 to for example, a power of two, z. B. 2 ¹⁶ specified and the features that are predetermined by their value in terms of crashes are normalized to this range of values. In this way, functions that process these characteristics for the evaluation of the activation of the personal protection devices can always count on only one value range and one

Adjustment with respect to different value ranges is not necessary.

These at least two mapped features then enter into the activation algorithm, which is characterized, for example, by time-dependent and / or time-independent thresholds. It is possible that others too

Sizes as the characteristics affect the height of the thresholds or characteristics. Various classification methods can also be used here.

Mainly under a control unit for controlling the

Personal protection means understood an electrical device which processes the at least one sensor signal and generates a drive signal for controlling the personal protection means in dependence thereon. The control unit accordingly has corresponding analog and / or digital computing means.

In the present case, it also has at least one interface which provides the at least one sensor signal of the accident sensor system. The accident sensor system can be arranged inside and / or outside of the control unit. Ie. The interface can be hardware-wise, for example, as part of an ASIC, but also in software and as a software module, for example, on a

Microcontroller be formed.

The software and / or hardware configuration can also for the analysis circuit for generating the at least two features of the at least one sensor signal for the mapping circuit for mapping the at least two features on the single value range and for the drive circuit for generating the drive signal in response to the at least two mapped features apply. Ie. these circuits may be software modules on a microcontroller, but they may each be arranged as hardware circuits, for example on an ASIC be. A mixture of hardware and software design is possible in the present case.

By the measures and refinements recited in the dependent claims are advantageous improvements of the independent

Claims specified method or control device for controlling personal protection means for a vehicle possible.

It is advantageous that the imaging takes place by a value-wise division of the at least two features. The value-wise division means that the at least two features which are present as numbers, for example in the microcontroller, are mapped by a corresponding division to the single value range. The simplest division is a halving, which can preferably be done by a so-called bit shift. However, other divisions are possible. Halving has the advantage that it can be made very simple, but it also involves a loss of information.

Alternatively, it is advantageously possible to perform the mapping by a so-called nominator-denominator method. In this case, the at least two features are each multiplied by a predetermined fraction. This is a special kind of division. The denominator of this fraction can advantageously be designed as a power of two. A power of two can currently be performed on a microcontroller or other computer about 80 times faster than a division. An example of such

Denominator is ^2.sup.5 = 32. The numerator of this fraction may advantageously be determined as a function of a factor which multiplies a maximum value of the respective feature to the next power of two. This means that the measured maximum value of a feature is multiplied to the next higher power of two. This factor then determines the

Counter, wherein the numerator can then be an integer number, for example, by multiplying this factor by the denominator, which is a power of two. With the multiplication of the feature with this break thus formed one is at the next higher power of two and a division, for example by 2, the final normalization or mapping takes place on the single value range.

A further advantageous embodiment is that before mapping the at least two features are increased or also upscaled.

This can be, for example, around 10%, wherein it is advantageous in this scaling up that real crashes are also imaged, because the training data, which are used in the laboratory, especially in machine-learning-based methods, can be exceeded by real crashes in the size of the features , In addition, by scaling up

Deviations in the sensor signal e.g. due to temperature drift, installation inaccuracies or aging.

Advantageously, depending on an evaluation of the at least two features, an exchange of the at least two features and / or a

Power on and / or off functions for these features. By mapping to the single value range, the interchangeability of the features has become simple, so that if it turns out that a functional module does not make a sufficient contribution to the algorithm performance, this module can be easily exchanged without having to adjust the value range of the feature , This is illustrated by the following examples:

Example 1: If, for example, the vehicle front structure changes significantly in the vehicle development process, this has effects on the crash characteristics. Recently, the tendency to shorten the front structure with simultaneous stiffening observed. This new structure tends to better accommodate features that account for high frequencies in the acceleration signal. The adaptation to this new feature is easier with the described exchange.

Example 2:

During the development process of a vehicle, it may happen, for example, that a sensor, for example an upfront sensor, is saved. Features that detect offset crashes by upfront sensor pairs are then ineffective. These Features can then be replaced by better features (eg obtained from yaw rate sensors for offset detection) without significantly affecting the software architecture.

Example 3:

Functionality can be specifically added or removed in the course of the crash: If, for example, a filling frontal crash can be ruled out after a certain crash duration, functionality could be added which detects slower crashes, for example with a small overlap or central or non-central pile crashes. The latter crash types can also be triggered later. This example is covered by the patent application DE102007004345A1 by Hiemer and Kolatschek. The present patent application simplifies the flow control described there with time-controlled and / or event-controlled switching off and adding functionality substantially.

Embodiments of the invention are illustrated in the drawings and are explained in more detail in the following description.

Show it

1 shows a block diagram of the control device according to the invention with connected components in a vehicle,

FIG. 2 shows a first flow chart,

3 shows a first block diagram, FIG. 4 shows a second block diagram,

FIG. 5 shows a register,

FIG. 6 shows a feature time diagram and

FIG. 7 shows a second flowchart.

FIG. 1 shows a block diagram of the control unit SG according to the invention in a vehicle FZ. Components are connected to the control unit in order to ensure the function of this control unit for the activation of personal protective equipment. In the present case, only the components necessary for understanding the invention are shown. Other necessary for the operation of the control unit SG components such as a power supply are for the sake of simplicity has been omitted. However, the formation of these components is known to the skilled person from the prior art.

An air pressure sensor PPSr is connected to an interface IF1 in the control unit SG. This air pressure sensor PPSr is located in the side part of the vehicle, on the right side. This air pressure sensor PPSr is used to detect a side impact. For the sake of simplicity, only the air pressure sensor system on the right side of the vehicle is shown here. In reality, a further air pressure sensor system will also be arranged on the left side, but for the sake of simplicity this has also been omitted here. This also applies to the acceleration sensor system PASr, which is connected to the interface IF2 in the control unit SG and is located, for example, in the region of the B pillar of the vehicle FZ. Furthermore, a so-called sensor cluster DCU is connected to the control unit SG, which has a wide variety of sensors, such as acceleration sensors, for example.

Rotation rate sensors and / or structure-borne sound sensors. This sensor cluster DCU is connected to the interface I F3. The interfaces I Fl, I F2 and I F3 can be configured separately as hardware components or also on an ASIC as individual sections on this ASIC. It is also possible to display at least parts or some of these interfaces as software modules. In the present case, only a section of the available sensors is shown. It can be connected to the control unit SG more sensors and also sensors are located within the control unit SG.

The interfaces I Fl, I F2 and I F3 are connected to a microcontroller μC in the control unit SG, for example via the so-called SPI (Serial Peripheral Interface Bus) connected. The SPI bus is a serial bus consisting of 5 parallel lines, a first line for data transmission from a slave to a master (MISO), a second line for data transmission from the master to the slave (MOSI). , a third line for selecting the slave (CS), an enable line and a clock line are provided. The microcontroller .mu.C carries out the method according to the invention and, in dependence on the sequence of this method, generates a drive signal, which is also transmitted, for example, via the SPI bus to a drive circuit FLIC which activates or activates the personal protection means PS, for example by energizing. The drive circuit

FLIC can also be part of a system ASIC, as well as the interfaces I Fl to 3.

As already stated above, other components of the control device SG as well as a separate evaluation path have been omitted for the sake of simplicity.

FIG. 2 explains in a first flowchart the sequence of the method according to the invention. In step 200, the interfaces IF1 to 3 provide the at least one sensor signal, so that the

Microcontroller μC via the SPI bus, these signals preferably digitally read and process.

The microcontroller .mu.C initially executes the generation of the at least two features from the at least one sensor signal in method step 201. In this case, the microcontroller .mu.C or another component which carries out the generation of the at least two features, for example an acceleration signal as characteristics, a filtered acceleration signal, an integrated acceleration, a twofold integrated acceleration, a derivative of the acceleration, a maximum value acceleration, a

Evaluate the mean value of the acceleration or other signal characteristics as characteristics. Accordingly, he can do so for the air pressure signal of the air pressure sensor PPSr, wherein from the air pressure signal, for example, the integrated air pressure signal, etc. can be generated as features. Signal signals such as the so-called closing velocity, ie the impact velocity, time to impact, ie the time to impact and the other parameter as features, can also be obtained from signals from an environmental sensor system. In method step 202, the at least two features are then mapped to the single value range. This can be done, as shown above, by a simple division or the so-called nominator-denominator method. But other imaging techniques may be used herein. For example, truncating or adding the value range of

Characteristics or also a cutting out of the value range of the feature, which is then further used as a value range. The formation of the activation decision then ensues in method step 203 with these depicted features. Depending on this activation decision, the personal protection means are activated, as indicated above.

FIG. 3 illustrates in a block diagram how the part of the control device according to the invention runs in detail. The at least one sensor signal 300 is provided by the interface I F4, namely the microcontroller .mu.C, which, for example as a software module, has the analysis circuit AS which generates the features M1 and M2 from the at least one sensor signal 300. These features M1 and M2 are input to the mapping circuit ABS, which performs the mapping on the at least one range of values as stated above. The features Ml * and M2 * so depicted are supplied to the drive circuit ANS, which supplies the drive signal 301 in FIG

Dependent on these mapped features Ml * and M2 * performs. In the present case, the two features are exemplary only. The inventive method or the control unit according to the invention are particularly suitable for a larger number of features.

FIG. 4 shows this part of the control unit in conjunction with a specific activation algorithm 403. However, the invention can also be part of the activation algorithm itself. The at least one sensor signal a, in this case the acceleration signal, enters the analysis circuit AS, such as the at least two features Ml and M2, for example, the forward displacement and the speed reduction by dual or simple integration generated as features. The mapping circuit ABS then performs the mapping of these two features Ml and M2 on the single value range and thus generates the features Ml * and M2 *, which enter the influencing circuit ANSB, which is part of the drive circuit ANS. These Influencing circuit ANSB generates an influencing signal 405 which influences a characteristic 402 in the activation algorithm 403. Namely, this characteristic curve 402 separates the triggering cases 401 from the non-triggering cases 400 in an acceleration-speed reduction diagram. In this diagram, value pairs of the acceleration a and of the deceleration dV enter into this diagram.

These value pairs are then compared with the characteristic curve 402 with respect to their position and it is determined whether or not these value pairs indicate a triggering event. Therefore, it is clear that the location of the map 402 determines whether or not a certain pair of values indicates a driving case for the airbags. If a control case is identified, then the

Control algorithm 403 from the drive signal 404, which then leads to the activation of the personal protection means.

Figure 5 shows a bit register 57 with successive bit positions 50 to 56, where 50 is the most significant bit and 56 the least significant one. Indicated by a simple bit shift by the arrows, succeeds a division of a feature and thus an image on a single range of values. This is a particularly simple method, but also causes a loss of information in the respective feature by the image.

A better quality with respect to the classification has the so-called nominator-denominator method, which is explained in Figure 6 using a signal timing diagram. The characteristic curve is shown here by the reference numeral 603. The characteristic curve is limited to the signal range (= normalized), which is determined by the maximum value 605.

A particularly resource conserving method for mapping the signal 603 into the range described by the maximum value 605 is to apply a simple bit shift to the right of the waveform 603. This results in the signal described by the designation 601. The bracket 600 shows the area lost by this figure. A loss of signal information is therefore given here, but the image itself is carried out in the simplest manner and thus already the resources of the microcontroller or other computing means in the control unit SG. In contrast to the simple bit shift, a more precise method with significantly less loss of signal information is given when the signal profile of the feature 603 is mapped onto the signal profile 602. Then the target value range, which is limited by the value 605 upwards, much better used. In order to map the signal 603 to the signal 602, the signal 603 is first scaled up, for example by 10%, for the reasons stated above, from which the signal profile 604 results. Advantageously, the scaled-up waveform 604 is then further processed by the nominator-denominator method. First, the upscaled signal 604 is multiplied by the nominator to, for example, next higher

Power of 2, which is described by the designation 606. Subsequently, the signal is imaged onto the target profile 602 on the basis of the signal path which is highly multiplied up to 606 by means of division by the denominator. Advantageously, the denominator is a power of 2. In this case, the division on the microcontroller or other computing means in the control unit SG cost can be mapped by a shift operation.

By applying the nominator-denominator method, the range of values limited by 604 is thus utilized to the greatest possible extent.

In the event that the maximum value of the signal 603 is below the target value range 605, the described process is completely analog. Also in this case, the signal curve is optimally adapted to the value range specified by 605.

The nominator-denominator method thus works for normalization "up" if the target range is greater than the maximum value of the feature.Nominator-denominator method also works for normalization "down" if the target range is less than of the

Maximum value of the feature.

FIG. 7 shows the method according to the invention in a further flow chart. In method step 700, the determination of the maximum feature value for all selected features during the calibration carried out. In method step 701, the determination of the counter and the denominator is carried out by means of a so-called look-up table. The numerator and denominator values are stored, for example, in the EEPROM, specifically in method step 702. Steps 700 to 702 are carried out on the laboratory, while the following steps 703 and 704 take place in the control unit. In this case, the feature values are multiplied by the fraction in method step 703, and in method step 704, the classification and thus the decision as to whether the personal protection means are to be activated or not take place with these mapped features.

Claims

claims

1. A method for controlling personal protection devices (PS) for a vehicle (FZ) with the following method steps:

Providing at least one sensor signal (300) of an accident sensor system (DCU, PPSr, PASr),

Generating at least two features (M1, M2) from the at least one sensor signal (300), characterized by the mapping of the at least two features (M1, M2) to a single value range, - activating the personal protection means (PS) in dependence on the at least two featured features (Ml *, M2 *).

2. The method according to claim 1, characterized in that the mapping is carried out by a value division of the at least two features (Ml, M2).

3. The method according to claim 2, characterized in that the value division takes place by a bit shift.

4. The method according to claim 1 or 2, characterized in that the

Mapping is performed by a nominator-denominator method in which the at least two features (Ml, M2) are each multiplied by a predetermined fraction.

5. The method according to claim 4, characterized in that a power of two is used as the denominator of the fraction.

6. The method according to claim 4 or 5, characterized in that the counter of the fracture is determined in dependence on a factor, wherein the Factor multiplied a maximum value of the respective feature to the next higher power of two.

7. The method according to claim 6, characterized in that the counter is formed as an integer.

8. The method according to any one of the preceding claims, characterized in that the at least two features are increased before imaging.

9. The method according to any one of the preceding claims, characterized in that in response to an evaluation of the at least two features an exchange of the at least two features and / or a connection and / or a shutdown of functions takes place.

10. Control unit (SG) for controlling personal protection means (PS) for a vehicle (FZ), comprising: an interface (IF1 to 4) which provides at least one sensor signal (300) to an accident sensor system (DCU, PPSr, PASr), - an analysis circuit (AS), which generates at least two features (M1, M2) from the at least one sensor signal (300), characterized in that an imaging circuit (ABS) is provided in the control unit (SG), which comprises the at least two features (M1, M2). maps to a single value range and that a drive circuit (ANS) is provided which generates a drive signal for driving the personal protection means (PS) in dependence on the at least two mapped features (Ml *, M2 *).