DETECTION OF AN ABNORMAL SIGNAL IN A COMPOUND SAMPLED SIGNAL

• US 20100106441A1
• Filed: 11/10/2009
• Published: 04/29/2010
• Est. Priority Date: 05/11/2007
• Status: Active Grant

• Current Assignee: Service Solutions US LLC (Robert Bosch GmbH)
• Original Assignee: Teradyne Diagnostic Solutions Limited (SPX Corporation)
• Inventors: SHI, Zhanqun, KIDD, Moray
• Granted Patent: US 8,326,557 B2
• Time in Patent Office: Days
• US Class Current: 702/71
• CPC Class Codes: G01H 3/08 Analysing frequencies present in complex vibrations, e.g. comparing harmonics present acoustic presence detection G01V1/001
• CPC Class Codes: G01R 31/08 Locating faults in cables, transmission lines, or networks
• CPC Class Codes: G06F 15/00 Digital computers in general details G06F1/00 – G06F13/00; Data processing equipment in general

Abstract

A technique detects an abnormal signal in a compound sampled signal recorded in the time domain. The technique involves dividing the sampled signal recorded in the time domain into sample segments; transforming each of the sample segments from the time domain into the frequency domain to determine transformed segments, each transformed segment having frequency points, each frequency point having an amplitude associated with a certain frequency; for a frequency point in a given transformed segment, determining a ratio by dividing the amplitude of the frequency point by a value indicative of an average of the amplitudes of the frequency points at the same frequency across transformed segments; repeating the determination of a ratio for frequency points in each transformed segment to determine ratios for each transformed segment; repeating the determination of ratios for transformed segments; and using the ratios to detect the abnormal signal in the compound sampled signal.

Claims

1. A method of detecting an abnormal signal in a compound sampled signal recorded in the time domain, the method comprising:

dividing the sampled signal recorded in the time domain into a plurality of sample segments;

transforming each of the plurality of the sample segments from the time domain into the frequency domain to determine a plurality of transformed sample segments, each transformed sample segment having a plurality of frequency points, each frequency point having an amplitude associated with a certain frequency;

for a frequency point in a given transformed sample segment, determining a ratio by dividing the amplitude of the frequency point by a value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments;

repeating the determination of a ratio for a plurality of frequency points in each transformed sample segment to determine a plurality of ratios for each transformed sample segment;

repeating the determination of a plurality of ratios for a plurality of transformed sample segments; and

using the ratios to detect the abnormal signal in the compound sampled signal.

2. The method of claim 1, wherein a ratio is determined for a majority of the frequency points of each transformed sample segment.

3. The method of claim 2, wherein a ratio is determined for all of the frequency points of each transformed sample segment.

4. The method of claim 1, wherein the determination of a plurality of ratios is undertaken for a majority of the transformed sample segments.

5. The method of claim 1, wherein the determination of a plurality of ratios is undertaken for all of the transformed sample segments.

6. The method of claim 1, wherein the value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments is determined across a majority of transformed sample segments.

7. The method of claim 1, wherein the value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments is determined across all of the transformed sample segments.

8. The method of claim 1, wherein the value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments is indicative of the mean, median or mode value.

9. The method of claim 1, further comprising adding the plurality of ratios together to determine a summed ratio for each transformed sample segment.

10. The method of claim 9, wherein each ratio of the plurality is squared before they are added together to determine a summed squared ratio for each transformed sample segment.

11. The method of claim 9, wherein detecting the abnormal signal comprises identifying the sample segments having the highest summed ratios or summed squared ratios.

12. The method of claim 11, wherein the summed ratios or summed squared ratios are deemed to be the highest if they exceed a predetermined threshold.

13. The method of claim 12, wherein the threshold is a percentage of a peak summed ratio or summed squared ratio.

14. The method of claim 13, wherein the threshold is 60% of the peak summed ratio summed squared ratio.

15. The method of claim 11, further comprising taking away from the frequency points of a transformed sample segment deemed to have a highest summed ratio or summed squared ratio the frequency points of a transformed sample segment deemed not to have a highest summed ratio or summed squared ratio.

16. The method of claim 11, wherein the transformed sample segment or segments having the highest summed ratios or summed squared ratios are transformed from the frequency domain into the time domain.

17. The method of claim 11, wherein a location of transformed sample segments in the frequency domain having the highest summed ratios or summed squared ratios is used to identify the corresponding sample segments in the time domain.

18. The method of claim 16, wherein the sample segment in the time domain is analysed to determine if the sample segment in the time domain contains some or all of the abnormal signal.

19. The method of claim 18, wherein the analysis comprises playing the signal back to a human.

20. The method of claim 18, wherein the analysis comprises using detection hardware or software to compare the sample segments in the time domain with a plurality of stored signals.

21. A computer readable medium carrying computer readable instructions configured to cause a computer to:

divide the sampled signal recorded in the time domain into a plurality of sample segments;

transform each of the plurality of the sample segments from the time domain into the frequency domain to determine a plurality of transformed sample segments, each transformed sample segment having a plurality of frequency points, each frequency point having an amplitude associated with a certain frequency;

for a frequency point in a given transformed sample segment, determine a ratio by dividing the amplitude of the frequency point by a value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments;

repeat the determination of a ratio for a plurality of frequency points in each transformed sample segment to determine a plurality of ratios for each transformed sample segment;

repeat the determination of a plurality of ratios for a plurality of transformed sample segments; and

use the ratios to detect the abnormal signal in the compound sampled signal.

22. A computer apparatus for detecting an abnormal signal in a compound sampled signal, comprising:

a memory storing processor readable instructions; and

a processor for reading and executing instructions stored in the memory;

wherein the processor readable instructions include instructions configured to cause the processor to;

divide the sampled signal recorded in the time domain into a plurality of sample segments;

transform each of the plurality of the sample segments from the time domain into the frequency domain to determine a plurality of transformed sample segments, each transformed sample segment having a plurality of frequency points, each frequency point having an amplitude associated with a certain frequency;

for a frequency point in a given transformed sample segment, determine a ratio by dividing the amplitude of the frequency point by a value indicative of an average of the amplitudes of the frequency points at the same frequency across a plurality of transformed sample segments;

repeat the determination of a ratio for a plurality of frequency points in each transformed sample segment to determine a plurality of ratios for each transformed sample segment;

repeat the determination of a plurality of ratios for a plurality of transformed sample segments; and

use the ratios to detect the abnormal signal in the compound sampled signal.