Method for detecting the application of scratches and bumps

Abstract

A method for detecting scratches and bumps applied to a vehicle uses an acceleration sensor. Signals of the sensor are analyzed and evaluated based on duration and frequency.

Claims

1. A method for detecting the application of scratches and/or bumps to a vehicle using at least one acceleration sensor mounted to the vehicle, wherein signals which are output by the acceleration sensor are monitored and analyzed by an electronic device, the method comprising: analyzing an envelope of the signals; determining that a signal is to be evaluated based on analyzing the envelope; and evaluating the signal to classify an application of a scratch and/or bump, wherein the evaluating is on the basis of a duration and a frequency of the signal in each of a plurality of frequency bands which are respectively filtered out from a total bandwidth of the signal, wherein the evaluating includes evaluating an amplitude of the signal in each of the plurality of frequency bands to classify the application of the scratch and/or bump based on a relationship between the amplitudes.

2. A method according to claim 1, wherein the at least one acceleration sensor senses an acceleration in three perpendicular axes and the plurality of frequency bands is filtered out and evaluated from signals of all three axes.

3. A method according to claim 1, wherein the signals are evaluated in a frequency band having a total width of about 0-4000 Hz.

4. A method according to claim 1, wherein the at least one acceleration sensor does not directly contact an impact surface of the vehicle.

5. A method according to claim 1, wherein analyzing the envelope of the signals comprises setting an amplitude threshold and filtering out signals having an amplitude that exceeds the threshold.

6. A method according to claim 1, wherein evaluating the signal in the plurality of frequency bands comprises extracting an upper envelope of the signal by extracting values above an average value of the signal.

7. A method according to claim 1, wherein evaluating the signal comprises determining a decay percentage of an amplitude of the signal in each frequency band.

8. A method according to claim 1, wherein evaluating the signal comprises determining a damping characteristic of the signal in each frequency band.

9. A method according to claim 1, wherein the signal is classified into categories based on a combination of a time threshold and a square of the time threshold, said categories comprising a scratch, multiple bumps and single bumps with or without damage.

10. A method according to claim 1, wherein the signal is classified into categories based on an amplitude threshold over a median of the signal, said categories comprising scratches and multiple bumps.

11. A method according to claim 1, wherein the signal is classified into categories based on an average of peak amplitudes, a decay percentage of an amplitude and a damping characteristic of the signal, said categories comprising single bumps with or without damage.

12. An electronic device configured to carry out the method of claim 1.

13. A non-transitory computer readable medium comprising instructions for carrying out the method of claim 1.

14. A method according to claim 1, wherein the plurality of frequency bands have a width in a range of 150 Hz to 250 Hz.

Description

DRAWINGS

(1) Exemplary embodiments and functions of the present disclosure are described herein in conjunction with the following drawings, showing schematically in:

(2) FIG. 1—a flowchart illustrating an envelope analysis;

(3) FIG. 2—characteristics of a bump;

(4) FIG. 3—characteristics of a scratch;

(5) FIG. 4—a signal classified into the category no event;

(6) FIG. 5—a signal classified into the category scratch;

(7) FIG. 6—a signal classified into the category bump without damage;

(8) FIG. 7—a signal classified into the category bump with damage; and

(9) FIG. 8—a flowchart illustrating an event classification.

DETAILED DESCRIPTION

(10) FIG. 1 illustrates various steps of the disclosed method performed within a system for determining if a signal is to be evaluated or not. The data received from at least one acceleration sensor, e.g. data from an acceleration sensor sensing the acceleration in three perpendicular axes, are first imported at 10. Thereafter, it is determined if a signal is to be evaluated or not based on an analysis of the envelope of the signals. This may comprise to set an amplitude threshold and to filter out signals having an amplitude that exceed the threshold.

(11) In a next step a filter design 12 is applied to filter out several discrete frequency bands from a total bandwidth of the signal. In an exemplary method a total bandwidth of 0 to 1600 Hz is filtered and divided into eight frequency windows or frequency sub-bands each having a bandwidth of e.g. 200 Hz (see also FIGS. 4 to 7). As a result, the filter 12 outputs twenty-four signals resulting from three axes and eight frequency sub-bands.

(12) The analysis of the envelope of the signal includes a determination of an event occurrence time at step 18 said event occurrence time including a raise time 20 and a decay time 22 of the signal. Alternatively, the analysis of the envelope of the signal may include to set an amplitude threshold and to filter out signals having an amplitude that exceeds the threshold.

(13) FIG. 2 illustrates an example of a signal resulting from a bump with damage, said signal being retrieved from a frequency sub-band of 1200 to 1400 Hz. As shown in FIG. 2, the raise time 20 and the decay time 22 show about the same duration of respectively about 0.1 s. In contrast thereto, FIG. 3 shows a signal resulting from a scratch filtered out in a frequency band of 1200 to 1400 Hz wherein the raise time (up to a maximum value) is about 1 s whereas the decay time 22 of the signal has a much higher duration of about 6 s.

(14) As shown in FIG. 1, for determining the decay time 22 of the signal a peak value 26 resulting from a maximum value and a median value of the signal is determined (see also FIGS. 2 and 3) and on the basis of the resulting raise time and decay time the event occurrence time is determined in step 24. If the event occurrence time lies above a preset threshold, an event occurrence is set at step 28 and the signal data is input into a classifier, the operating principle of this classifier being shown in FIG. 8. The preset level of a threshold is determined in accordance with a parameter set for use during runtime of the present system. Such parameters may be stored in memory of the system, e.g., nonvolatile memory, and the threshold can be used to adapt the sensitivity of the present system based on the stored parameters.

(15) FIG. 8 shows that the raw data is imported at 10 to receive at 30 data which is processed in FIG. 1 to determine if a signal is to be evaluated or not. If the event occurrence time 28 or an amplitude lies below a preset threshold it is decided in step 32 that the category of “no event” applies and the method returns to step 10. If the event occurrence time 28 or an amplitude lies above the threshold it is determined that the signal is further to be evaluated and the signal will be evaluated starting in step 34 and based on a duration and a frequency of the signal. First, it is determined if a long-time event 36 or a short-time event 38 is analyzed. If a long-time event 36 is analyzed, a scratch (FIG. 3) or multiple bumps (FIG. 2) could have been applied to the vehicle. If a short-time event 38 (FIG. 2) has been detected, an analysis will follow if the signal shows a relatively low amplitude in higher frequencies resulting from a single bump 44 or if relatively high amplitudes in higher frequency ranges are detected resulting from a bump 46 causing a damage. Number ranges for the relatively lower and higher amplitudes of the different classes of events are usually not fixed values. Rather, those numbers are calculated in dependence of the event, the vehicle and of customer requirements (what event has to be detected). Those values (coefficients) will be calculated offline based on an initial analysis during development/validation of the respective vehicle. Those calculated coefficients are used during run-time as a combination with input values for a neural network function. Based on this combination different outputs are received as different events.

(16) After it has been detected in step 42 that multiple bumps were analyzed, it is further analyzed if a long-time event signal shows a lower amplitude in higher frequencies resulting from multiple bumps 48 without causing a damage. Alternatively, if the signal shows a higher amplitude in higher frequencies, multiple bumps 50 causing a damage have been applied to the vehicle.

(17) FIGS. 4 to 7 show signals of different categories allowing the classification described above. FIG. 4 shows very small amplitudes in all frequency sub-bands. Accordingly, the resulting signal is classified as “no event” according to step 32 of FIG. 8.

(18) FIG. 5 shows a long-time event 36, namely a signal having a duration of several seconds and a relatively high amplitude over several seconds in a higher frequency range of about 1000 to 1600 Hz. Accordingly, this event is classified as a scratch 40.

(19) FIG. 6 shows a short-time event 38 with lower amplitudes in lower frequencies (the maximum is at about 400 Hz) reflecting a signal bump event 44.

(20) Finally, FIG. 7 shows very high amplitudes with a relatively short duration in low and in high frequency ranges resulting from a bump 46 causing a damage.

(21) The identification of these apparent relationships between the relatively lower and higher amplitudes and the lower and higher frequencies provide, at least in part, for the relatively accurate event classification of the present disclosure. The specific values of these relationships for the event classification may be (dynamically) set depending on the events to be classified and the vehicle or vehicle type. Typically, these values (e.g., coefficients) are calculated based on an initial analysis regarding the vehicles. Those calculated coefficients are then used during runtime of the present system as a combination with input values for a neural network function. Based on this combination the different outputs of the system provide for the classification of the different events.

(22) The present disclosure includes the following embodiments:

First Embodiment

(23) A method for detecting the application of scratches and/or bumps to a vehicle using at least one acceleration sensor mounted to the vehicle, wherein signals which are output by the acceleration sensor are monitored and analyzed by means of an electronic device, the method including: analyzing the envelope of the signals, determining if a signal is to be evaluated based on the envelope analysis; and—if the signal is to be evaluated, evaluating the signal on the basis of a duration and a frequency of the signal.

(24) A method according to the first embodiment, wherein analyzing the envelope of the signal includes a determination of an event occurrence time which includes a rise time and a decay time of the signal.

(25) A method according to the above embodiment, wherein the signals are classified into a long-time event and a short-time event based on the event occurrence time.

(26) A method according to the first embodiment, wherein the signals are classified into categories, said categories including no event, scratch, bump without damage and bump with damage.

(27) A method according to as above, wherein the long-time events are classified into scratches and multiple bumps based on the sequence of the amplitude of the signal over time.

(28) A method as above, wherein the bumps are classified into bumps with and without damage based on an amplitude and a frequency of the signal.

(29) A method as above, wherein the short-time events are classified into single bumps with and without damage based on an amplitude and a frequency of the signal.

(30) A method according to the first embodiment, wherein the signals are evaluated in a plurality of frequency bands, in particular in eight frequency bands, which are respectively filtered out from the total bandwidth of the signal.

(31) A method as above, wherein the frequency bands have a width of about 150-250 Hz, e.g. 200 Hz.

(32) A method according to the first embodiment, wherein the signals are evaluated in a frequency band having a total width of about 1400-1800 Hz, e.g. 1600 Hz.

(33) A method according to the first embodiment, wherein the sensor senses an acceleration in three axes.

Second Embodiment

(34) A method for detecting the application of scratches and/or bumps to a vehicle using at least one acceleration sensor mounted to the vehicle, wherein signals which are output by the acceleration sensor are monitored and analyzed by means of an electronic device, the method including: analyzing the envelope of the signals, determining if a signal is to be evaluated based on the envelope analysis; and—if the signal is to be evaluated—evaluating the signal on the basis of a duration and a frequency of the signal, wherein the signals are evaluated in a plurality of frequency bands, of e.g. 150-250 Hz, which are respectively filtered out from the total bandwidth of the signal.

(35) A method according to the second embodiment, wherein the sensor senses an acceleration in three perpendicular axes and the plurality of frequency bands is filtered out and evaluated from signals of all three axes.

(36) A method according to the second embodiment, wherein the signals are evaluated in a frequency band having a total width of about 0-4000 Hz, e.g. 0-1600 Hz or 0-800 Hz.

(37) A method according to the second embodiment, wherein an acceleration sensor is used that does not directly contact an impact surface of the vehicle.

(38) A method according to the second embodiment, wherein analyzing the envelope of the signals includes setting an amplitude threshold and filtering out signals having an amplitude that exceeds the threshold.

(39) A method according to the second embodiment, wherein the evaluation of the signals in the plurality of frequency bands includes extracting an upper envelope of a signal, e.g. extracting values above an average value of the signal.

(40) A method according to the second embodiment, wherein evaluating the signals includes a determination of a decay percentage of an amplitude of a signal in each frequency band.

(41) A method according to the second embodiment, wherein evaluating the signals includes a determination of a damping characteristic of a signal in each frequency band.

(42) A method according to the second embodiment, wherein the signals are classified into categories based on a combination of a time threshold and the square of the time threshold, said categories including a scratch, multiple bumps and single bumps with or without damage.

(43) A method according to the second embodiment, wherein the signals are classified into categories based on an amplitude threshold over a median of the signal, said categories including scratches and multiple bumps.

(44) A method according to the second embodiment, wherein the signals are classified into categories based on an average of peak amplitudes, a decay percentage of an amplitude and a damping characteristic of a signal, said categories including single bumps with or without damage.

(45) An electronic device configured to carry out the method of the first or the second embodiment.

(46) A non-transitory computer readable medium including instructions for carrying out the method of the first or the second embodiment.