Device and process for signal analysis

• US 6,591,682 B1
• Filed: 08/14/2000
• Issued: 07/15/2003
• Est. Priority Date: 08/14/2000
• Status: Expired due to Fees

• Current Assignee: PRÜFTECHNIK Dieter Busch AG (Fortive Corp.)
• Original Assignee: PRÜFTECHNIK Dieter Busch AG (Fortive Corp.)
• Inventors: Lysen, Heinrich
• Primary Examiner(s): Kwok, Helen
• Assistant Examiner(s): SAINT SURIN, JACQUES M
• Application Number:
  US09/637,581
• Time in Patent Office: 1,065 Days
• Field of Search: 736/60, 736/02, 735/93, 736/59, 735/87, 735/79, 735/97, 735/98, 736/46, 702/35, 702/56
• US Class Current: 73/602
• CPC Class Codes: G01H 1/003 of rotating machines G01H1/10 takes precedence
• CPC Class Codes: G01M 13/045 Acoustic or vibration analysis
• CPC Class Codes: G01N 2291/014 Resonance or resonant frequency
• CPC Class Codes: G01N 2291/048 Transmission, i.e. analysed material between transmitter and receiver
• CPC Class Codes: G01N 2291/2626 Wires, bars, rods
• CPC Class Codes: G01N 2291/2634 cylindrical from outside
• CPC Class Codes: G01N 2291/2696 Wheels, Gears, Bearings
• CPC Class Codes: G01N 29/07 by measuring propagation velocity or propagation time of acoustic waves
• CPC Class Codes: G01N 29/11 by measuring attenuation of acoustic waves
• CPC Class Codes: G01N 29/14 using acoustic emission techniques echo of particles G01N29/046; measuring mechanical vibrations or acoustic waves in solids in general G01H1/00
• CPC Class Codes: G01N 29/4427 with stored values, e.g. threshold values
• CPC Class Codes: G01N 29/4436 with a reference signal amplitude comparison G01N29/48
• CPC Class Codes: G01N 29/46 by spectral analysis, e.g. Fourier analysis or wavelet analysis spectral signal processing per se G06F17/14

Abstract

A device for detecting or analyzing machinery damage is used preferably in the determination of defects in roller bearings. Pulse-like machinery or bearing noise acquired by a sensor is converted into special time signals. These time signals correspond to events which are basically periodic in occurrence. The time signals are sent to one, preferably several, classification units depending on the time differences. Before classification, combination of adjacent time signals or events is performed using their pulse heights. The combining is multiplicative or obeys some other bivalent relation. Classification forms a frequency distribution with an abscissa which is divided into time units. Specific characteristics of the frequency distribution provide information on incipient or manifest machinery damage.

Claims

1. Device for signal analysis comprising:

a sampling and conversion means for forming sample values of a signal, and for digitizing the sample values;

a selection unit for selecting specific sample values;

an arithmetic-logic operation unit having means for combining a pair of selected specific sample values according to a defined relation and for outputting resultant values; and

classification means connected to receive the resultant values from the arithmetic-logic operation unit for classifying the resultant values and producing a distribution of the resultant values, which distribution is based on distances that separate each combined pair of sample values;

an evaluation unit for evaluating a distribution of said resultant values obtained from the classification means; and

,

computational means for determining a significant frequency value, an integral multiple of a time difference value assigned to the significant frequency value, and for determining frequency values of the integral multiple of time difference values; and

wherein the evaluation unit has an evaluation means for the frequency values.

2. Device according to claim 1, wherein said means for combining has means for producing a summation of sample values;

and wherein said distribution produced by the classification means is a frequency distribution of the values formed by summation.

3. Device as claimed in claim 1, wherein said evaluation unit evaluates the classification results for detection or analysis of machinery damage.

4. Device as claimed in claim 3, wherein said evaluation unit is functionally connected to a comparison unit for evaluating the classification results for detection of roller bearing defects.

5. Device for detection or for analysis of machinery damage or defects which cause noise, with determined and stochastic properties, comprising:

a sensor for acquiring machinery noise and for conversion of said noise into an electrical signal;

a rectifier for forming an absolute value of the electrical signal;

at least one peak value detector; and

at least one time analyzer for analysis of output signals of the at least one peak value detector for determining of special events, defined by time and intensity, from said electrical signal, wherein the at least one time analyzer has means for determining respective instants of intermediate maximum and minimum values of a signal function as time events; and

wherein a buffer storage is provided for storing the respective instants together with the respective signal amplitudes.

6. Device according to claim 5, wherein said at least one peak detector comprises at least two peak detectors, each peak detector having a different recovery time or decay time constant.

7. Device as claimed in claim 5, wherein a multiplier is provided for producing product values from stored respective instants representing time events at predetermined time differences and signal amplitudes of the signal function;

and wherein means for subjecting the product values to frequency analysis is provided.

8. Device as claimed in claim 7, further comprising an unit for acquisition and proportional subtraction of time values which correspond to a multiple of a base time value.

9. Device as claimed in claim 7, further comprising an evaluation unit for evaluating the frequency analysis and for determining whether, what type, and to what extent machinery damage is reflected in the electrical signal.

10. Device as claimed in claim 5, further comprising a registration unit for storing selected time and intensity values over a longer time interval than provided by said buffer storage for enabling long term recording and analysis of machinery component damage.

11. Device as claimed in claim 5, comprising a logic operation unit for producing an event value from two event intensities at a time by one of product forming, summing, and maximum value determining.

12. Device as claimed in claim 5, wherein a means for at least one preprocessing, rectification and absolute-value generation of said electrical signal is provided.

13. Device as claimed in claim 5, further comprising evaluation means for forming and evaluating several frequency distributions in a multiplicity of damage duration categories.

14. System for detecting repeated, pulse-like signal portions of a time signal comprising

a sensor for acquiring and converting a time signal into an electrical AC signal;

an electronic sampling means for sampling the AC signal and for generating a sampled, time discrete signal;

at least one reference signal generator for generating sawtooth reference signals with a steep rise and comparatively slow fall;

at least one trigger stage for repetitive triggering of the reference signal generators;

a signal analysis stage which uses scanned, time-discrete signals for triggering the at least one trigger stage and which has means for comparing the sampled, time-discrete signals to the sawtooth reference signals;

a storage unit for storing special sample values with respect to an instant of their occurrence and an amplitude thereof;

an arithmetic unit for forming arithmetic products;

a classification unit for forming at least one histogram; and

a pattern recognition unit for classification of histograms according to predetermined criteria.

15. System according to claim 14, further comprising a signal conversion means for producing positive definite signal portions, and at least one filter means.

16. System as claimed in claim 14, further comprising an analysis unit for determining a number of events and sending a number of events combined with one another for histogram analysis until an event or an evaluation of the histogram analysis is determined to reflect a steady-state condition.

17. System as claimed in claim 14, further comprising an analysis unit for determining a number of events and sending a number of events combined with one another for histogram analysis until an event or an evaluation of the histogram analysis is determined to be statistically relevant according to predetermined criteria.

18. System as claimed in claim 14, further comprising at least one comparison stage for checking that predetermined limit values have been exceeded, have not been reached, or have been equaled;

and a signaling stage for outputting a signal if values filed in the histogram exceed the predetermined limit values.

19. System as claimed in claim 14, further comprising a comparator stage and a digital arithmetic unit for generating time-variable threshold signals which are sent to the comparator stage for use as comparison signals for determining time events.

20. Process for signal analysis comprising the steps of

using a sampling and conversion means for forming sample values of a signal;

digitizing the signal;

combining each sample value in a logic operation unit, which combines each sample value of a signal with a plurality of the sample values of the signal, using a logic operation performed according to one of product formation, sum formation, and maximum formation of at least two relative maximum values of the signal;

sending the results of the logic operation for classification by means of a classification unit, and performing classification in the manner of a distribution; and

using an evaluation unit for evaluation of one of a classification result and a distribution of frequency distribution, wherein said evaluation is performed using the further steps of determining a significant frequency value, determining an integral multiple of a time difference value which is assigned to the significant frequency value, determining frequency values at integral multiple of time difference values; and

evaluating the frequency values determined.

21. Process as claimed in claim 20, further comprising the step of determining the existence of machinery damage from said evaluation.

22. Process as claimed in claim 21, wherein the machinery damage determined is the existence of roller bearing defects.