Method and device for detecting and locating a fault point of a housing
Patent Information
- Application Number
- IL308271
- Authority / Receiving Office
- IL · IL
- Patent Type
- Patents
- Current Assignee / Owner
- Priority Date
- 2021-05-07
- Filing Date
- 2022-05-03
- Publication Date
- 2026-08-01
- Estimated Expiration
- 2042-05-03
AI Technical Summary
Existing acoustic imperfection detection methods are unable to precisely and reliably localize defects in device housings, particularly in security-critical applications, as they often fail to differentiate between subtle changes in sound transmission patterns caused by manipulations or damage.
A method involving the arrangement of multiple loudspeakers and microphones within a housing to measure reference and current sound transmission frequency responses, calculating differential responses, and using correlation measures to identify and localize disturbances by determining the combination with the highest correlation, particularly effective in the 250 Hz to 2 kHz frequency band.
This approach allows for reliable and precise localization of defects, achieving high spatial resolution and increased reliability through the use of correlation measures and threshold values, effectively identifying and pinpointing disturbances within the housing.
Smart Images
Figure 00000024_0000 
Figure 00000024_0001 
Figure 00000025_0000
Abstract
Description
[0001] Method and device for detecting and locating a fault in a housing. The present invention relates to methods and devices for detecting and locating faults in a housing using sound waves. Monitoring the integrity of device housings is of strategic importance for security-critical applications in the government, industrial, or military sectors, particularly for electronic devices such as servers, computers, laptops, notebooks, and smartphones, as the "first line of defense" against hacker attacks. Tampering with a housing, for example, by drilling, inserting foreign objects such as probes or transmitters, or destroying seals inside the housing, can be an unmistakable sign of such hacker attacks. With a large number of device housings to be monitored, e.g., in server cabinets or in the field, or if they are not detectable from the outside, such physical tampering remainsHowever, they often go undetected. Damage during field use, which may lead to leaks, for example, often remains hidden. The use of sound waves in the form of structure-borne or airborne sound is known for monitoring the integrity of structures. For example, US 2017 / 0338804 A1 describes distributing several loudspeaker / microphone pairs throughout an interior space to be monitored. The airborne sound transmission path of each loudspeaker / microphone pair is simulated in a connected evaluation circuit as an adaptive digital filter. If the adaptive digital filter changes too significantly compared to an initial state, a disturbance in the interior space is detected and localized as being near the affected loudspeaker / microphone pair. However, the known acoustic disturbance detection methods are not capable of locating the detected disturbances more precisely and reliably, so that they can be quickly located.The invention aims to create fault detection methods and devices that are capable of reliably locating the detected faults. This aim is achieved in a first aspect with a method for detecting and locating a fault in a housing, in particular for an electronic device, comprising: arranging at least two loudspeakers and at least two microphones in the interior of the housing; for each possible pairing of loudspeaker and microphone: measuring a reference frequency response of the sound transmission from the loudspeaker to the microphone of this pairing; at a later time, for each pairing: measuring a current frequency response of the sound transmission from the loudspeaker to the microphone of this pairing; for each pairing: determining a difference frequency response as the difference between the current frequency response of this pairing and the reference frequency response of this pairing; for at least onePairing: Determining a detection measure of the difference frequency response of this pairing and, if this at least one detection measure fulfills a predetermined detection criterion: Detecting an interference point, for each possible combination of two pairings: Calculating a correlation measure between the difference frequency responses of the pairings of this combination, Determining the combination with the highest correlation measure and, if the same loudspeaker or microphone occurs in the two pairings of this combination, Localizing the interference point as closer to this loudspeaker or microphone than to any other loudspeaker or microphone. The method according to the invention allows for reliable localization of an interference point to the surrounding area of one of the loudspeakers or microphones. If a correspondingly high number of loudspeakers or microphones is distributed largely evenly over the interior of the housing, a correspondingly high spatial resolution of theLocalization can be achieved. Any correlation measure can be used as a correlation measure for the agreement of two difference frequency responses, for example a sum of the (frequency-wise) absolute or squared differences or an L1, L2, or Lp norm of the differences between the two difference frequency responses considered, a scalar product of these difference frequency responses, etc. The correlation measure is preferably determined from a cross-correlation of the difference frequency responses in a selected frequency band, which allows reliable localization. It is particularly advantageous if the frequency band is 250 Hz to 2 kHz, preferably 350 Hz to 1.1 kHz. Evaluation in these frequency bands has proven to be a particularly reliable localization criterion. In a further preferred embodiment of the invention, when determining the combination with the highest correlation measure, only those combinations are considered that have aexceed a correlation threshold. This can further increase the reliability of the localization. The detection criterion for detecting a disturbance could, for example, be the maximum amplitude of the difference frequency responses of all pairs, if this exceeds a threshold. Preferably, the detection criterion is met if the total signal energy of at least one difference frequency response exceeds a detection threshold. The reference frequency response and the current frequency response of the sound transmission between the loudspeaker and microphone of a pair can be measured in a variety of ways. In a first embodiment of the invention, the respective frequency response is measured using chirps emitted by the loudspeakers. In a second embodiment, the respective frequency response can be measured using Dirac pulses emitted by the loudspeakers. The measurement using chirps takes somewhat longer, but leads tomore accurate results, and the sound power received by the microphone and recorded over time directly reflects the frequency response when the chirp is tuned across the frequency - e.g., linearly or logarithmically. Measurement using Dirac pulses is faster, but requires a subsequent Fourier transformation to calculate the frequency response from the impulse response recorded by the microphone over time. In each of the embodiments mentioned, the reliability of the localizations can be further increased if the reference frequency responses and current frequency responses are each averaged over several individual measurements, for example, over two or more immediately consecutive individual measurements. In a second aspect, the invention provides a device for detecting and locating a fault in a housing, in particular for an electronic device, comprising: at least two loudspeakers and at least twoMicrophones for arrangement in the interior of the housing; and a signal generator connected to the loudspeakers and an evaluation circuit connected to the microphones; wherein the evaluation circuit is designed to measure, for each possible pairing of loudspeaker and microphone, a reference frequency response of the sound transmission from the loudspeaker to the microphone of this pairing, at a later time, for each pairing: to measure a current frequency response of the sound transmission from the loudspeaker to the microphone of this pairing, for each pairing: to determine a difference frequency response as the difference between the current frequency response of this pairing and the reference frequency response of this pairing, for at least one pairing: to determine a detection measure of the difference frequency response of this pairing and, if this at least one detection measure fulfills a predetermined detection criterion: to detect an interference point and for each possible combination of twoPairings: to calculate a correlation measure between the difference frequency responses of this combination, and then to determine the combination with the highest correlation measure and, if the same loudspeaker or microphone occurs in the two pairings of this combination, to localize the disturbance as being closer to this loudspeaker or microphone than to any other loudspeaker or microphone. Regarding the advantages and preferred embodiments of the device according to the invention, reference is made to the above explanations regarding the method carried out by the device. The invention is explained in more detail below with reference to exemplary embodiments illustrated in the attached drawings. The drawings show: Fig. 1 a block diagram of the device of the invention installed in a device housing to be monitored for carrying out the method of the invention; Fig. 2 a flow diagram of the method of the invention; Fig. 3 an example of aReference frequency response and a current frequency response in an amplitude / frequency diagram; Fig. 4 shows an example of a difference frequency response for a housing without a defect in an amplitude / frequency diagram; Fig. 5 shows an example of a difference frequency response for a housing with a defect in an amplitude / frequency diagram; Fig. 6 shows an example of correlation measures of difference frequency responses from six possible combinations of four possible pairs of loudspeakers and microphones in a device according to the invention with two loudspeakers and two microphones. Fig. 1 shows a device 1 which is also suitable for carrying out the method shown in Fig. 2. The device 1 is used to monitor the integrity of a housing 2 in order to detect and classify defects F therein. The housing 2 has an interior 3 for accommodating technical components 4. For example, the housing 2 is the housing of a computer, server, notebook,Laptops, smartphones, etc. The housing 2 is not necessarily closed. The interior 3 is filled with a gas for sound propagation, usually just ambient air, which is either pre-filled or penetrates into the interior 3 through openings in the housing 2. The disturbance points F for which the housing 2 is to be monitored can either be defects, i.e. points where something is “missing” in or on the housing 2, such as a hole, a crack or break in the wall of the housing 2, a technical component 4 removed from the housing 2, a removed or broken seal, or the like. Or the disturbance point F is a foreign body, i.e. something additional in the housing 2, for example a probe, a tap wire, a transmitter, or the like. If a component 4 changes its position in the housing, e.g. becomes loose due to an impact and shifts, then a defect occurs at its original location and it formsa foreign body at its new location. For the detection and localization of interference points F, the device 1 comprises two or more loudspeakers L arranged in the interior 3 of the housing 2. i (i = 1, 2, …, I) and - spaced from these - two or more microphones M j (j = 1, 2, …, J). It is advantageous to use the loudspeakers L i and especially the microphones M j as evenly as possible over the interior 3, since the localization of an interference point F later in the spatial resolution of the respective surrounding areas of the microphones M j The number I of loudspeakers L i and the number J of microphones M j does not necessarily have to be the same. For each possible pairing (i, j) from a loudspeaker L i and a microphone M j can a sound transmission S i,j The sound transmission S i,j(“acoustic transmission path”) includes both the direct sound from the loudspeaker L i to the microphone M j as well as any form of indirect sound that is caused by reflections on the inner walls of the housing 2 or on components 4 in the interior 3 of the housing 2 from the respective loudspeaker L i to the respective microphone M j the pairing (i, j). In order to capture as much indirect sound as possible and thus cover the interior 3 extensively, the loudspeakers L i and / or microphones M j be omnidirectional or have correspondingly wide directional characteristics. Each speaker L i is connected to the output of a signal generator 5 and each microphone M jconnected to the input of an evaluation circuit 6. The signal generator 5 and the evaluation circuit 6 can communicate with each other for the purpose of synchronization. The signal generator 5 can be realized together with the evaluation circuit 6, for example, as a microprocessor or implemented on such a microprocessor using software. The signal generator 5 and the evaluation circuit 6 can also be realized by already existing technical components 4 of the device in the housing 2. With reference to Fig. 2, with the aid of the device 1, in a first step 7, on a "healthy", ie non-manipulated housing 2 without defects F, for each possible pairing (i, j) of a loudspeaker L i and a microphone M j a reference frequency response REF i,j the sound transmission S i,j between this speaker L i and microphone M jmeasured. For the sake of simplicity, the set {(i, j)} of all possible pairings (i, j) is numbered with the index m (m = 1, 2, ..., M), ie a mapping function Φ between the pairings (i, j) and the indices m = Φ(i, j) is chosen. In principle, the mapping function Φ can be chosen arbitrarily, as long as it is unique and bijective. Examples of the mapping function Φ are m = i ∙ (J - 1) + j or m = i + j ∙ (I - 1). For example, in the case of two loudspeakers L1, L2 and two microphones M1, M2, there are four possible pairings (i, j), namely (1, 1), (2, 1), (1, 2), (2, 2), and these are numbered with the indices m = 1, 2, 3, 4, for example, as follows: The reference frequency response REF i,j a pairing (i, j) with the index m is therefore referred to as REF m be referred to, see Fig. 3. For the measurement of the reference frequency response REF m the sound transmission S i,jThere are several possibilities. On the one hand, the signal generator 5 can generate a chirp that is emitted by the respective loudspeaker L i radiated, i.e. a monofrequency signal whose frequency is tuned over time across the frequency range to be measured. The signals picked up by the respective microphone M j The recorded sound amplitude or power is recorded over time and immediately provides the reference frequency response REF m over the frequency. Or the signal generator 5 generates a Dirac pulse, which the respective loudspeaker L i radiates, and the respective microphone M j The impulse response recorded over time is then Fourier transformed to obtain the reference frequency response REF mover the frequency. The emitted signal could also be a coded signal, e.g., it is "chopped" again with a higher-frequency signal (e.g., 40 kHz), thus creating a pseudo-noise signal. The measured reference frequency responses REF m are stored for later fault location monitoring, for example in a memory of the evaluation unit 6. The reference frequency responses REF m For example, they are determined in the audible acoustic frequency range of 20 Hz - 20 kHz, but can also - alternatively or additionally - include the infrasound range below 20 Hz or the ultrasonic range above 20 kHz, e.g., up to 100 kHz or 200 kHz. The choice of frequency range is determined, among other things, by the housing dimensions at the lower end and the size of detectable defects at the upper end. It is understood that in step 7, the reference frequency responses REF mare measured one after the other to determine the sound transmission distances S i,j to be able to measure acoustically cleanly and separately from each other. Optionally, each reference frequency response REF m from several individual measurements, ie, using several consecutive chirps or Dirac pulses. It is also possible to average each reference frequency response REF m to be filtered and / or weighted before further use, for example to filter out impact noise below 200 Hz, to smooth out outliers or notches, or to filter out frequency ranges in which the received signal of a microphone M j provides a signal-to-noise ratio that is too low for meaningful evaluation. In order to detect the occurrence of a disturbance F and then to localize it, a current frequency response SPEC is calculated for each pairing (i, j) or m = Φ(i, j) at any later time in a step 8.m (Fig. 3) of the sound transmission S i,j between the respective loudspeaker L i and the respective microphone M j In measurement step 8, the same measurement method is used that was used for measuring the reference frequency responses REF m used in step 7 to determine the current frequency response SPEC m with the reference frequency responses REF m to make them comparable. When measuring 8 of the current frequency responses SPEC m (optional) averaging over several individual measurements or (optional) filtering, weighting, or smoothing can be omitted to save computing time. The current frequency responses SPEC m are measured one after the other in step 8 to measure the sound transmissions S i,j of the pairings to be acoustically cleanly separated from each other. After measuring the current frequency response SPEC mIn step 9, for each pairing m = Φ(i, j) a difference frequency response DIF m (Fig. 4, 5) from the deviation of the respective current frequency response SPEC m from the respective reference frequency response REF, ie the same pairing m m Each difference frequency response DIF m For example, a simple (frequency-wise) difference (subtraction) between the current frequency response SPEC m and the reference frequency response REF m of the pairing m or another deviation measure, for example the (frequency-wise) absolute or squared differences or the like between the two curves REF m and SPEC m . In an optional step 10, each difference frequency response DIF m weighted or filtered if desired. Subsequently, in step 11, a global measure V m (Fig. 4, 5) at least one or each difference frequency response DIF mcalculated and checked in step 12 against a detection criterion to detect the presence of a defect F. The measure V m For example, the maximum amplitude value of the respective difference frequency response DIF m be an average of this difference frequency response DIF m , after it has been subjected to an absolute value formation (absolute value formation), or the total “signal energy” E m of the respective difference frequency response DIF m The latter can be calculated, for example, by calculating the difference frequency response DIF mis first subjected to absolute value formation and then integrated over the entire frequency range of interest, e.g., 20 Hz to 20 kHz. Steps 8 - 11 are performed for each pairing m = Φ(i, j), see loop 13. Each of steps 8 - 11 could alternatively have its own loop 13. The comparison step 12 then proceeds. The comparison and thus detection criterion of step 12 is met if, for example, at least one of the - or all - measures V m exceeds or exceeds a detection threshold D. It is understood that in step 12, several measures V m can be evaluated together. For example, a sum or an average of all measures V mbe compared with the detection threshold D. If the detection criterion of the comparison step 12 is not met (branch "n"), then no disturbance F was detected and the method returns - if necessary after expiration of a waiting time ΔT in step 14 - in a loop 15 to the first step 8 of the loop 13 for a renewed measurement of the current frequency responses SPEC mback. The waiting time ΔT can, for example, be a few seconds or minutes, so that the housing 2 in the loop 15 is periodically monitored at these intervals. If, on the other hand, the comparison step 12 shows that the detection criterion is met (branch "y"), the presence of a disturbance source F has been detected and the process proceeds to the localization section 16 - 19. The detection threshold D can be selected empirically and optionally also adaptively adjusted using automatic control: In the error-free case, i.e., without a disturbance F, it can, for example, be reduced to such an extent that even small changes to the housing 2, i.e., the occurrence of small disturbances F, can be detected. In the localization section 16 - 19, in a first step 16, a correlation measure R is calculated for each possible combination (m, n) of two pairings (m, n). m,n between the difference frequency responses DIF m and DIF nof the pairings of this combination (m, n) is calculated. The index n denotes a different pairing from the set {m = 1, 2, M} than the index m, ie n ǂ m. Furthermore, for symmetric correlation measures V m R m,n = R n,m , so that for M pairings there are a total of possible
[0002] Combinations and thus correlation measures R m / n The step
[0003] 16 is therefore executed 17 times in the loop, ie for all m, n for which m > n.
[0004] Fig. 6 shows for a device 1 with two loudspeakers L1, L2 and two microphones M1, M2 and thus M = 4 possible
[0005] Pairings m or n all possible combinations for which each have a correlation measure R m,n the respective difference frequency responses DIF m and DIF n was calculated, see Table 2 below:
[0006] The correlation measure Rm,n between two difference frequency responses DIF m , DIF n can be calculated in any way known in the art, as long as it is a measure of the deviation between the two difference frequency responses, for example summed absolute or squared differences, an L1, L2 or Lp norm of a difference function DIF m - DIF n , a scalar product of the difference frequency responses DIF m , DIF n , etc. In the example shown, the correlation measure R m,n from a cross-correlation of the two considered difference frequency responses DIF m and DIF n determined, in particular from the cross-correlation for τ = 0, either over the entire frequency range f or optionally in a selected frequency band. The selected frequency band is, for example, 250 Hz to 2 kHz, in particular 350 Hz to 1.1 kHz. Subsequently, in step 18, the combination (m, n) is determined which yields the highest correlation measure R m,n of all combinations. In the example of Fig. 6, this is the first combination (1, 2). So far, all reference frequency responses REF m , current frequency response SPEC m and difference frequency responses DIF m measured and evaluated in real terms with regard to their magnitudes (amplitudes), ie, without taking their phases into account; optionally, however, they could also be measured and evaluated in complex terms. In this case, R can be used as the correlation measure. m,n For the evaluation in step 18, for example, the amount of cross-correlation at τ = 0, where * denotes the complex conjugation. When searching for the highest correlation measure in step 18, only those correlation measures that exceed a correlation threshold C (Fig. 6) can optionally be considered. The correlation threshold C can be determined empirically. With appropriate normalization of the correlation measure R m,nFor the range -1 … +1 (see Fig. 6), the correlation threshold C is, for example, in the range from 0.5 to 1, in particular at approximately 0.75. In subsequent comparison steps 19 and 20, the loudspeaker and microphone indices i, j of those two pairings m, n that form the combination (m, n) determined in step 18 are examined. For this purpose, the first comparison step 19 examines whether the same loudspeaker index i occurs in the two pairings m, n of the combination determined in step 18, and the second comparison step 20 examines whether the same microphone index j occurs in these two pairings m, n. If the comparison step 19 results in "yes" (branch "y"), the disturbance F is considered to be in the vicinity of the loudspeaker L i localized, and if the comparison step 20 results in “yes” (branch “y”), then the disturbance F is considered to be in the neighborhood of the microphone M jlocalized. It is understood that the order of comparison steps 19 and 20 is arbitrary. The term “neighborhood” of loudspeaker L i or microphone M j It is understood that the disturbance point F is assigned to this loudspeaker L i or microphone M j closer than any other speaker L k≠i or microphone M k≠j If both comparison steps 19 and 20 result in "no" (branches "n"), then the localization is ambiguous and the method optionally returns—and possibly after the waiting time ΔT in step 14 has elapsed—to the first step 8 of loop 13. The invention is not limited to the illustrated embodiments, but encompasses all variants, modifications, and combinations thereof that fall within the scope of the appended claims.
Claims
Claims:
1. Method for detecting and locating a defect (F) of a housing (2), in particular for an electronic device, comprising: arranging at least two loudspeakers (L i ) and at least two microphones (M j ) inside the housing (2); characterized by: for each possible pairing (m) of loudspeaker (L i ) and microphone (M j ): Measuring a reference frequency response (REF m ) sound transmission (S i,j ) from the loudspeaker (L i ) to the microphone (M j ) of this pairing (m); at a later time, for each pairing (m): measuring a current frequency response (SPEC) m ) sound transmission (S i,j ) from the loudspeaker (L i ) to the microphone (M j ) of this pairing (m); for each pairing (m): determining a difference frequency response (DIF) m ) as the difference between the current frequency response (SPEC m) this pairing (m) and the reference frequency response (REF m ) this pairing (m); for at least one pairing (m): determining a detection measure (V m ) of the difference frequency response (DIF) m ) this pairing and, if this is at least one detection measure (V m ) a predefined detection criterion is met: Detecting a defect (F), for each possible combination (m, n) of two pairings: Calculating a correlation measure (R) m,n ) between the Differential frequency responses (DIF) m , DIF n ) of the pairings of this combination, determining the combination (m, n) with the highest correlation coefficient (R) m,n ) and, if the same loudspeaker (L) is used in both pairings of this combination i ) or the same microphone (M j ) occurs, localizing the fault (F) as this loudspeaker (L) i ) or microphone (M j) closer than any other loudspeaker or microphone.
2. Method according to claim 1, characterized in that the correlation measure (R) m,n ) from a cross-correlation of the difference frequency responses (DIF) m , DIF n ) is determined in a selected frequency band.
3. Method according to claim 2, characterized in that the frequency band is 250 Hz to 2 kHz, preferably 350 Hz to 1.1 kHz.
4. Method according to any one of claims 1 to 3, characterized in that, when determining the combination (m, n) with the highest correlation coefficient (R), m,n ) only those combinations are considered that exceed a correlation threshold (C).
5. Method according to one of claims 1 to 4, characterized in that the detection criterion is fulfilled when the total signal energy of at least one differential frequency response (DIF) m) exceeds a detection threshold (D).
6. Method according to one of claims 1 to 5, characterized in that the respective frequency response (DEF) m , SPEC m ) by means of the loudspeakers (L i ) emitted chirps are measured.
7. Method according to one of claims 1 to 6, characterized in that the respective frequency response (DEF) m , SPEC m ) by means of the loudspeakers (L i ) emitted Dirac pulses are measured.
8. Method according to one of claims 1 to 7, characterized in that the reference frequency responses (REF m ) and current frequency responses (SPEC) m ) are averaged over several individual measurements.
9. Device for locating a fault (F) in a housing (2), in particular for an electronic device, comprising: at least two loudspeakers (L i ) and at least two microphones (M j) for arrangement in the interior (3) of the housing (2); and one connected to the loudspeakers (L i ) connected signal generator (5) and one connected to the microphones (M j ) connected evaluation circuit (6); characterized in that the evaluation circuit (6) is designed to perform, for each possible pairing (m) of loudspeaker (L i ) and microphone (M j ) a reference frequency response (REF m ) sound transmission (S i,j ) from the loudspeaker (L i ) to the microphone (M j ) to measure this pairing (m), at a later time, for each pairing (m): a current frequency response (SPEC) m ) sound transmission (S i,j ) from the Speaker (L i ) to the microphone (M j ) of this pairing (m) to measure, for each pairing (m): a difference frequency response (DIF) m ) as the difference between the current frequency response (SPEC m) this pairing and the reference frequency response (REF m ) to determine this pairing, for at least one pairing (m): a detection measure (V m ) of the difference frequency response (DIF) m ) to determine this pairing and, if this at least provides a detection measure (V) m ) fulfills a predefined detection criterion: to detect a defect (F) and for each possible combination (m, n) of two pairings: a correlation measure (R) m,n ) between the difference frequency responses (DIF) m , DIF n ) to calculate this combination, and then the combination (m, n) with the highest correlation measure (R) m,n ) to determine and, if the same loudspeaker (L) is used in the two pairings (m, n) of this combination i ) or the same microphone (M j ) occurs, the fault (F) as this loudspeaker (L) i ) or microphone (M j) to locate it closer than any other loudspeaker or microphone.
10. Device according to claim 9, characterized in that the correlation measure (R) m,n ) from a cross-correlation of the difference frequency responses (DIF) m , DIF n ) is determined in a selected frequency band.
11. Device according to claim 10, characterized in that the frequency band is 250 Hz to 2 kHz, preferably 350 Hz to 1.1 kHz.
12. Device according to any one of claims 9 to 11, characterized in that the signal generator (5) for each loudspeaker (L i ) generates chirps.
13. Device according to one of claims 9 to 11, characterized in that the signal generator (5) for each loudspeaker (L i ) Dirac impulses are generated.
14. Device according to one of claims 9 to 13, characterized in that the reference frequency responses (REF m ) and current frequency responses (SPEC) m) are averaged over several individual measurements.