Analysis of Oversampled High Frequency Vibration Signals

Abstract

A method of distinguishing a first physical phenomenon captured in a sensory measurement time waveform from a second physical phenomenon captured in the waveform includes: receiving the waveform on a processor from a sensor in sensory contact with an object undergoing first and second physical phenomena, wherein the first phenomenon is a comparatively fast event; deriving a first rate of change data stream from the time waveform with a processor operable on a processor, wherein each value of the first rate of change data stream is based on a difference in extreme amplitudes of the waveform during a first interval of waveform samples; and analyzing with the processor the derived first rate of change data stream to distinguish the comparatively fast first physical phenomena from the second physical phenomenon captured in the waveform.

Claims

1. A method of distinguishing a first physical phenomenon captured in a sensory measurement time waveform from a second physical phenomenon captured in the waveform, the method comprising: receiving the waveform on a processor from a sensor in sensory contact with an object undergoing first and second physical phenomena, wherein the first phenomenon is a comparatively fast event; deriving and pass filtering a first rate of change data stream from the time waveform with a processor, wherein each value of the pass filtered first rate of change data stream is based on a difference in extreme amplitudes of the waveform during a first quantity of waveform samples; deriving and pass filtering a second rate of change data stream from the time waveform with the processor, wherein each value of the pass filtered second rate of change data stream is based on a difference in extreme amplitudes of the waveform during a second quantity of waveform samples, and wherein the second quantity is larger than the first quantity of waveform samples; analyzing with the processor the derived first rate of change data stream and the second rate of change data stream to distinguish the comparatively fast first physical phenomenon from the second physical phenomenon captured in the waveform.

2. (canceled)

3. The method of claim 1, further comprising comparing the derived first rate of change data stream and the derived second rate of change data stream with the processor to further determine one or more of friction, shear, rubbing, stiction, and sliding characteristics of the first physical phenomenon.

4. The method of claim 1, further comprising applying expert logic to the derived first rate of change data stream, the expert logic selected from the group consisting of waveform segmentation logic, threshold logic, event type logic, deductive logic, knowledge based logic, metadata logic, application logic, observation logic, and action logic.

5. The method of claim 4, wherein the first and second phenomena are a machine friction event and vibration of a machine, and wherein the applied expert logic distinguishes a friction characteristic of the machine friction event.

6. The method of claim 5, wherein the applied expert logic characterizes the machine friction event as an event selected from the group consisting of negligible metal-to-metal contact, lubrication film breach, rolling contact, sliding contact, mixed mode lubrication regime, boundary lubrication regime, impact, rubbing, severe sliding, and adhesion.

7. The method of claim 5, further comprising: applying expert logic in a first step at a first location, the first step comprising a data compression step performed at the first location; and applying expert logic in a second step at a second location, the second step comprising a determination of one or more of findings, observations, recommendations, and knowledge based logic performed at the second location.

8. The method of claim 1, further comprising: deriving a decimated from the time waveform with the processor, the processor deriving the decimated stream by decimating the time waveform; and analyzing the decimated with the processor to further determine a characteristic from a list of the first and second physical phenomena.

9. The method of claim 1, wherein the first interval has a duration no greater than 10 μs.

10. The method of claim 1, wherein the first interval has a duration of no greater than 10 μs, and wherein the second interval has a duration of no greater than 500 μs.

11. The method of claim 1, further comprising segmenting the derived first rate of change data stream based on rate changes among sequenced values of the first rate of change data stream.

12. The method of claim 11, further comprising classifying each segment of the first rate of change data stream as one of an event and nonevent based on a value of the first rate of change data stream within each segment.

13. The method of claim 11, further comprising generating compressed data plot information corresponding to the segmented first rate of change data stream.

14. (canceled)

15. A method of distinguishing a friction event captured in a sensory time waveform, the method comprising: receiving the waveform on a processor from a sensor in sensory contact with an object of interest undergoing the friction event; deriving a pass filtered first rate of change data stream from the time waveform with a processor operable on the processor, wherein each sequential value of the first rate of change data stream is based on a rate of change of the time waveform during a first quantity of sample intervals; deriving a pass filtered second rate of change data stream from the time waveform with the processor, wherein each sequential value of the second rate of change data stream is based on a rate of change of the time waveform during a second interval that is longer than the first interval; analyzing with the processor the derived first rate of change data stream and the derived second rate of change data stream to determine characteristics of the friction event based on the rate of change of the time waveform over the first interval and the second interval.

16. The method of claim 11, further comprising selectively segmenting at least one of the first rate of change data stream and the second rate of change data streams based on one or more of abrupt increases and abrupt decreases in change rates in at least one of the first and second rate of change data streams.

17. The method of claim 11, wherein the first interval has a duration of approximately 10 μs, and wherein the second interval has a shorter duration of approximately 500 μs.

18. A method of analyzing a sensory waveform, the method comprising: receiving the waveform derived from a sensor in sensory contact with an object of interest, the waveform data including a sensing of one or more of a fast event phenomenon, an intermediate event phenomenon, and a slow event phenomenon; translating the waveform into a first wavelength pass filtered rate of change data stream wherein the pass filter is one or more of a short wavelength, a medium wavelength, and a long wavelength; translating the waveform into a second wavelength pass filtered rate of change data stream that is different from the first wavelength pass filtered rate of change data stream, wherein the pass filter is one or more of a short wavelength, a medium wavelength, and a long wavelength; and analyzing the one or more data streams by one or more of selective segmentation, selective decimation, threshold, segment association, event scoring, event classification, and deductive logic.

19. The method of claim 18, wherein the waveform data comprises a vibration waveform sampled at a rate of 102,400 samples per second, and wherein a fast phenomenon has a duration of approximately 0.04 ms.

20. The method of claim 18, wherein the rate of change waveform data is based on a difference in extreme amplitudes of the waveform data during a first interval of waveform samples.

21. The method of claim 20, wherein the short wavelength pass filter is less than approximately 0.01 ms, wherein the intermediate wavelength pass filter is less than approximately 0.05 ms, and wherein the long wavelength pass filter is less than approximately 0.5 ms.

22. (canceled)

23. (canceled)

24. (canceled)

25. (canceled)

26. (canceled)

27. (canceled)

28. (canceled)

29. (canceled)

30. (canceled)

31. The method of claim 1, wherein pass filtering the first rate of change data stream and the second rate of change data stream comprises a high frequency filter effect from a duration of peak-to-peak measurement intervals.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Further features, aspects, and advantages of the present disclosure will become better understood by reference to the following detailed description, appended claims, and accompanying figures, wherein elements are not to scale to more clearly show the details, wherein like reference numbers indicate like elements throughout the several views, and wherein:

[0037] FIG. 1 shows a sensor device according to one embodiment of the present disclosure;

[0038] FIG. 2 shows a flowchart of analyzation of a first rate of change data stream according to one embodiment of the present disclosure;

[0039] FIGS. 3 and 4 show exemplary first and second intervals of first and second rate of change data streams according to one embodiment of the present disclosure;

[0040] FIG. 5 shows a flowchart of analyzation of a first rate of change data stream and a second rate of change data stream according to one embodiment of the present disclosure;

[0041] FIG. 6 shows a visual icon of a segment of a rate of change data stream according to one embodiment of the present disclosure;

[0042] FIGS. 7A-7C illustrate segmentation of a rate of change data stream according to one embodiment of the present disclosure;

[0043] FIGS. 8 and 9 show characterization of event types based on segment data according to one embodiment of the present disclosure;

[0044] FIGS. 10A-10D illustrate an exemplary waveform, rate of change data stream plots, and compressed data plot corresponding to a decay event according to one embodiment of the present disclosure;

[0045] FIGS. 11A-11C show a compressed data plot based on oversampled waveform data according to one embodiment of the present disclosure;

[0046] FIG. 12 shows an output of compressed data plot information from a sensor device according to one embodiment of the present disclosure;

[0047] FIG. 13 shows a flowchart of generating compressed data plot information on a sensor device according to one embodiment of the present disclosure;

[0048] FIGS. 14A-14E illustrate analysis of a planar bearing failure according to one embodiment of the present disclosure;

[0049] FIGS. 15A-15E illustrate analysis of a roller bearing failure according to one embodiment of the present disclosure;

[0050] FIG. 16 shows an exemplary plot of multiple high pass filters performed on oversampled waveform data according to one embodiment of the present disclosure; and

[0051] FIG. 17 shows examples of lubrication boundary characterization according to one embodiment of the present disclosure.

DETAILED DESCRIPTION

[0052] Various terms used herein are intended to have particular meanings. Some of these terms are defined below for the purpose of clarity. The definitions given below are meant to cover all forms of the words being defined (e.g., singular, plural, present tense, past tense). If the definition of any term below diverges from the commonly understood and/or dictionary definition of such term, the definitions below control.

[0053] The present disclosure provides a method and apparatus for analyzing a waveform to distinguish a physical phenomenon that is a comparatively fast event. The method of the present disclosure distinguishes a first physical phenomenon, such as a vibration event on a machine, that is a comparatively fast event by analyzing a waveform of the event and characterizes the event based on the waveform of the fast event. The method for analyzing a waveform of the present disclosure enables automatic reporting of a fast event and data corresponding to the fast event while also compressing an amount of data transmitted by a device to provide information related to a fast event.

[0054] The method of the present disclosure preferably analyzes a waveform corresponding to an event on a machine or other object of interest. Referring to FIG. 1, the waveform is preferably generated using a sensor device 10 that is placed in sensory contact 13 with an object of interest. The sensor device includes a sensor 12 that responds to a physical stimulus on the object of interest such as by outputting a dynamic analog signal response, such as a 4 mA to a 20 mA or a 0V to 5V analog signal. As referred to in the present disclosure, sensory contact corresponds to the sensory device detecting physical stimuli of the object of interest including: mechanical vibration; airborne vibration; fluid borne vibration; stress wave; seismic wave; pressure wave; electric current; fatigue; fracture; slip; spallation; delamination; collision; abrasions; acoustic emission; adhesion; plastic deformation; cavitation; erosion; rubbing contact; severe sliding contact; stiction; and scintillation. Suitable sensor devices include, for example, the devices described in U.S. Pat. No. 9,389,610 and U.S. Patent Publication No. 2014/0121996, the contents of which are incorporated herein by reference in their entireties. An analog sensor output of the sensor on the sensor device is converted to digital waveform data, such as by an analog to digital converter (ADC). The analog sensor output is converted at a rate such that the converted digital waveform data is oversampled (such as between 51,200 and 204,800 samples per second). Oversampled data is also commonly referred to as high frequency data or high sampling rate.

[0055] The sensor device further preferably includes a housing 2 having a cover 1 and a sensor connection 4. The housing 2 with cover 1 and sensor connection 4 preferably contains a circuit 3 including an analog to digital converter (ADC) 14 in electronic communication with the sensor and a processor 16 in electronic communication with the analog to digital converter (ADC). A data storage unit 18 is in electronic communication with the processor. A power source 15 such as a battery or fuel cell or energy harvester or external power supply or combination preferably supplies electrical energy to sensor 12, processor 16, and communications connection 20. The data storage unit 18 may comprise either or both of persistent and volatile computer memory. Computer readable instructions are preferably stored on the data storage unit 18 and are operable on the processor 16. Module or software components are installed on the data storage unit 18 and are operable on the processor 16 to analyze the digital data waveform and distinguish a physical phenomenon that is a fast event, as discussed in greater detail below. Components of the sensor device 10 may be located on a single unit, such as a device that is in sensory contact with an object of interest. Additional components, such as additional central processing units and memory units, may be located remotely from the sensor device, such as on a remote diagnostic device. For example, a remote analysis device may be in wired or wireless communication with the sensor device 10 to receive and further analyze oversampled waveform data or other data output from the sensor device 10, such as via wireless communications connection 20.

[0056] Sensor device 10, or components thereof, are intended for use as part of a condition monitoring system. Preferred condition monitoring systems receive digital signals from the sensor device 10 and directly process those signals digitally for these systems, device 10 serves as a digital transducer which transforms typically analog sensor signals into machine readable digital information.

[0057] A first example digital transducer system is a remote analysis system which comprises a sensor device 10 in WiFi™ or Bluetooth™ or similar communication with a gateway which in turn communicates wirelessly by cellular communication with a remote analysis server having an internet portal interface for customer interaction. A system of this sort is described on WWW.WIRELESSVIBRATION.EU using trade names I-care, Wi-care, and I-see. This system in the art is benefitted by preferred embodiments of the sensor device 10 including these examples selective segmentation, selective decimation, signal compression, automatic tribological determinations regarding friction, lubrication and wear, automatic generation of observations, machine to machine transfer of low bandwidth high frequency analysis information. Battery life is a critically important factor for systems like this first example. To address battery life very large oversampled data streams are compressed using selective segmentation and selective decimation in the transducer so that small blobs of data such are preferably attached as metadata to each traditionally decimated waveform being transmitted from transducer to gateway.

[0058] A second example of the sensor device 10 being incorporated into a system as a digital transducer, is walk around data collection system similar to the Azima DLI Trio CX7 system, as described in http://azimadli.com/products/data-collectors/trio/, or similar data collection systems using either wired or wireless communication of digital measurement data to a handheld computer. In this second system there is plenty of electrical power bandwidth to enable preferred embodiments to transmit entire oversampled waveforms from the transducer to the handheld. In this situation the oversampled data may be further analyzed by traditional techniques such as FFT, synchronous time averaging, demodulation, etc., if an operator is skilled and desires to do those things in addition to fully automated steps of the present invention.

[0059] A third example of the sensor device 10 being incorporated into a system as a digital transducer, is to provide digital signal inputs to a control system or surveillance system. For control systems, it is common for a primary variable input to be an analog signal such as an overall for a dynamic signal; and it may be optional for secondary, tertiary, and quaternary variable inputs to transmit digital information. If a control system requires analog input, it is possible for digital signals of the sensor device to be converted from digital. There are numerous ways to incorporate outputs from a sensor device 10 into online monitoring systems ranging from substituting a pass filtered rate of change data streams in place of an unaltered time domain signals to adding time stamped event information chronological with to historian stored findings.

[0060] Data on the sensor device 10 is received on the sensor device, processed, and reported according to embodiments of the present disclosure. Data may be processed with either batched waveform chunks or continuous recursive flow of waveform data. Processes of the present disclosure may be persistent in memory of the sensor device 10, or may preferably not be retained on the data storage unit 18 of the sensor device 10. Instead, lossless data is retrieved and lossless reports transmitted from the sensor device 10, creating an output of suitable bandwidth for processing, transmission, and logistical analysis of the data.

Analysis of Oversampled Waveform Over One or More Intervals

[0061] Referring now to FIG. 2, methods of the present disclosure distinguish a first fast event captured in a time waveform. As referred to in the present disclosure, a fast event or fast event segment is an event within oversampled waveform data that is typically diminished, decayed, and even overlooked by existing data decimation practices, such as the system and method disclosed in U.S. Pat. No. 5,633,811. In contrast, a not fast event is a relatively long duration interval during which substantial information may be derived using traditional data decimation systems and methods such as those taught in U.S. Pat. No. 5,633,811. For example, a not fast event or combination of not fast event would include a one-times running speed high amplitude signal corresponding to an imbalance of a rotating machine wherein oversampled data may be decimated using known decimation practices without loss of substantial information about the imbalance. A specific example of a fast event is a stress wave signal pattern during which the signal rapidly and substantially increases within less than one common digital decimation practice interval. Fast is a subset of dynamic. In the two examples just listed above, imbalance and stress wave signal pattern are both dynamic.

[0062] In a first step 102, time waveform data such as an oversampled waveform data is received on a processor from a sensor device in sensory contact with an object of interest. The oversampled waveform data may include data related to a fast physical phenomenon transduced into a fast event within at least a portion of the oversampled waveform data. In step 104, a difference between a maximum amplitude and minimum amplitude of the oversampled waveform is determined within a first peak to peak interval. The peak to peak interval is a relatively short time interval or quantity of sequential sample intervals within the oversampled waveform data. In one preferable embodiment, a relatively long peak to peak interval has a duration of 500 μs or less. However, it is also understood that the first peak to peak interval may be less than a designated interval of 500 μs. A maximum amplitude and minimum amplitude of the oversampled waveform within an interval may occur adjacent to one another along the oversampled waveform over an interval that is substantially less than 500 μs, or the maximum amplitude and minimum amplitude of the oversampled waveform may occur at different points along the oversampled waveform at or near a full length of the interval.

[0063] A value corresponding to a difference in the maximum amplitude and minimum amplitude of the oversampled waveform data during the first interval is determined and stored on a memory unit. A series of values corresponding to a difference in the maximum amplitude and minimum amplitude of the oversampled waveform data during the first interval are determined and stored on a memory unit, with each sequential value derived from the difference in the maximum amplitude and minimum amplitude of the oversampled waveform data over consecutive first intervals. Each of the first intervals is an adjacent interval of the oversampled waveform data, and preferably each interval substantially overlaps an adjacent interval, such as by from about 5 μs to about 10 μs. The series of values corresponding to differences in the maximum amplitude and minimum amplitude of the oversampled waveform data are used to derive a first rate of change data stream, as shown in step 106.

[0064] Determination of a difference in a maximum amplitude and a minimum amplitude over the first interval of the oversampled waveform data may be applied to various measurements. For example, determination of a rate of change over an interval may include, but is not limited to, displacement or runout waveform data producing velocity information, velocity waveform data producing acceleration information, acceleration waveform data producing jerk information, and jerk waveform data producing hyperjerk information.

[0065] Referring now to FIGS. 3 and 4, exemplary diagrams show determination of a difference in a maximum amplitude and a minimum amplitude of an acceleration time waveform over a first interval 107 (also referred to as PP500 in the Figures). In the diagrams of both FIGS. 3 and 4, the first interval has a duration of 500 μs. In both the diagram of FIG. 3 and the diagram of FIG. 4, a value corresponding to the difference in the maximum amplitude and the minimum amplitude of the oversampled waveform is 10 g. In the diagram of FIG. 3, the oversampled waveform has approximately 26 peaks and 25 valleys over the duration of the first interval, and the value ranges from a maximum of 10 g to a minimum of 0 g. In contrast, the waveform data of FIG. 4 only includes one peak and one valley over the duration of the first 500 μs as interval, with a difference in the maximum amplitude and minimum amplitude over the first interval also being 10 g.

[0066] Deriving a difference in the maximum amplitude of the oversampled waveform and the minimum amplitude of the oversampled waveform data over a duration of the first relatively short interval performs a high pass filter of the oversampled waveform data to remove longer wavelength content of the oversampled waveform data and expose shorter wavelength information. The exposed pass filtered shorter wavelength information is then available for analysis to detect, quantify, and characterize fast events within the oversampled waveform data. For example, when the first interval has a duration of 10 μs, wavelengths shorter than 0.01 ms are passed and half-wavelengths of greater than 0.02 ms are filtered such that frequencies below 50,000 Hz are filtered while retaining higher frequency waveform data.

[0067] Referring again to FIG. 1, the derived first rate of change data stream of step 106 is subsequently analyzed to distinguish and characterize a type of event (such as impact, friction, and other types of events) at step 108. Expert logic, such as waveform segmentation logic, threshold logic, event type logic, deductive logic, knowledge based logic, metadata logic, application logic, and observation logic, is applied to the first rate of change data stream to provide additional characterization and analysis of the derived first rate of change data stream includes segmentation of the oversampled waveform data, flagging of events and nonevents, and compression of the oversampled data for transmittal and reporting, as discussed in greater detail below. Empirical information may be derived from logic for automatically interpreting rate of change data is derived and validated by experimental test results wherein physiological phenomena are in place and sensory data is collected and analyzed in a cause and effect manner. Empirical information may further be used through baseline measurement wherein acceptable levels of measure are derived from a baseline and departures above baseline are considered abnormal.

[0068] In one embodiment the method of distinguishing a first fast physical phenomenon related event captured in oversampled waveform data includes deriving a second rate of change data stream and further characterizing and analyzing a fast event of the oversampled waveform data based on both the first rate of change data stream and the second rate of change data stream, as shown in FIG. 5. Values of the second rate of change data stream are derived from a difference in a maximum amplitude and a minimum amplitude of the oversampled waveform data over a second interval, wherein the second interval has a duration that is longer than the first interval.

[0069] A series of values corresponding to a difference in the maximum amplitude and minimum amplitude of the oversampled waveform data during the second interval are determined and stored on the memory unit, with each value or each sequential value derived from the difference in the maximum amplitude and minimum amplitude of the oversampled waveform data over consecutive second intervals. Preferably each of the intervals is an adjacent interval of the oversampled waveform data, and preferably each interval overlaps an adjacent interval by from about 5 μs to about 10 μs. The series of values corresponding to differences in the maximum amplitude and minimum amplitude of the oversampled waveform data are used to derive the rate of change data stream.

[0070] For example, referring again to FIGS. 3 and 4, a second interval 109 (also referred to as PP10 in the Figures) may have a duration of approximately 10 μs. In the example of FIG. 3, a value corresponding to the difference in the maximum amplitude and the minimum amplitude of the oversampled waveform is 10 g, the same value as the derived value of the first interval 107. In the example of FIG. 4, a value corresponding to the difference in the maximum amplitude and the minimum amplitude of the oversampled waveform is 0.2 g.

[0071] Referring again to FIG. 5, in a first step 202 a fast event oversampled waveform is received from a sensor in sensory contact with an object of interest. At step 204, the difference in extreme amplitudes of the oversampled fast event waveform is determined during a first interval. A difference in extreme amplitudes of the fast event waveform over a second interval is further determined at step 205. A first rate of change data stream is derived based on the determined difference in extreme amplitudes of the fast event waveform during the first interval in step 206. A second rate of change data stream is similarly derived over the second interval at step 208. The derived second rate of change data stream of step 208 is subsequently analyzed to distinguish and characterize a type of event (such as impact, friction, and other types of events) at step 210. Expert logic is performed on the first rate of change data stream to provide additional characterization and analysis of the derived first rate of change data stream at step 212 including segmentation of the oversampled waveform data, flagging of events and nonevents, and compression of the oversampled data for transmittal and reporting, as discussed in greater detail below.

[0072] The first rate of change data stream and second rate of change data stream are analyzed to further characterize or determine a type of event captured in the oversampled data waveform. In the example shown in FIG. 3, a relatively large value corresponding to the difference in the maximum amplitude and minimum amplitude of the oversampled waveform over the second interval in relation to the first interval demonstrates a larger amount of acceleration change and smaller displacement in contrast with the example shown in FIG. 4. A relatively smaller value corresponding to the difference in the maximum amplitude and minimum amplitude of the oversampled waveform over the second interval in relation to the first interval demonstrates larger displacement. FIG. 3 shows values associated with an event such as a friction event, whereas FIG. 4 shows values associated with an event such as an impacting event involving a greater force moving over a greater distance.

[0073] When the second interval is an interval of 500 μs, the second interval passes signals having wavelengths shorter than 0.5 ms and filters signals having wavelengths of longer than 0.5 ms. Between 0.5 ms (corresponding to 2 kHz) and 1 ms (corresponding 1 kHz) half-waves may pass in an output of the first interval. Below 1 kHz even half waves are filtered. The first interval having a duration of approximately 10 μs passes signals having wavelengths shorter than 0.01 ms and filters signals having wavelengths longer than 0.01 ms. Between 0.01 ms (corresponding to 100 kHz) and 0.02 ms (corresponding to 50 kHz) half-waves may pass an output of the first interval. Below 50 kHz, even half waves are filtered.

[0074] The second, longer interval of 500 μs, filters out most signal content below 1,000 Hz and retains higher frequency rate of change content. The first, shorter interval of 10 μs, filters out most signal content below 50,000 Hz and retains higher frequency rate of change content. The first, shorter interval, is therefore a subset of the second longer interval and content of the first interval that is not within the second interval represents band pass rate of change content >1,000 Hz and <50,000 Hz.

[0075] In one embodiment, a third rate of change data stream is further derived from the oversampled waveform data for further analysis and characterization of a fast event captured in the waveform data. The third rate of change data stream may be based on a difference in a maximum amplitude and a minimum amplitude over a third interval, the third interval having yet another duration such as between the first and second intervals. The third rate of change data stream may alternatively correspond to additional rate of change values from the oversampled waveform data, such as hyperjerk values.

Selective Segmentation

[0076] Selected portions of the oversampled waveform data from the sensor device 10 may be further segmented based on analysis of the waveform data. An event, such as a machine event, may comprise one or more consecutive segments, wherein each segment may be defined by characteristics of the waveform data. Segments are selected based on the analyzed first rate of change data stream and second rate of change data stream described above. For example, a segment may be selected based on abrupt changes in values of the first rate of change data stream and second rate of change data stream, or when values exceed or fall below a desired threshold. Event segments represent all or part of an event, and may be either graphically represented on a display or mathematically represented in a time domain versus amplitude domain.

[0077] Each segment may be separated by a flag, wherein each flag is located at a point in the oversampled waveform data where the first rate of change data stream values abruptly increase or decrease. For example, a flag may be placed at each point along the oversample waveform data and the first and second rate of change data streams when the rate of change data streams have a substantial increase in value. Flags may designate the initiation of a fast event and termination of a fast event, such as by placement of flags when a rise or decrease in the first and second rate of change data streams is determined.

[0078] Referring to FIG. 6, each event segment visual representation includes left leg 25, right leg 27, body 23, neck 21, head 29 having amplitude in time T. The left leg 25 and right leg 27 represent minimum to average times of a flag indicating an event or nonevent. The body 23 represents an average value during an interval between flags. A width of the body 23 may be interpreted as a function of characteristics of the oversampled waveform data, such as relating inversely to a frequency of the data. The neck 21 extends from the average value of the body 23 to a maximum value of the head 29. A magnitude of the maximum value of the head 29 relative to the average value of the body 23 may be analyzed to interpret symmetry and variance with similarity and differences to ways skewness, kurtosis, standard deviation, and similar population metrics that may be applied.

[0079] FIG. 7A-7C shows information associated with a compressed data plot for a friction type event. FIG. 7A displays 50 compressed data plot, 52 segment max value, 54 friction type event, 56 statistical threshold, and 58 non-event. FIG. 7B displays 60 two rate of change plots (PP500 and PP10 ), 62 long wavelength (0.5 ms sample interval) rate of change data stream, and 64 short wavelength (0.01 ms sample interval) rate of change data stream. FIG. 7C displays 70 waveform data plot, and 72 oversampled time waveform. Table 1 shows automatic scoring associated with the compressed data plot of FIG. 7A. Note that automatic segmentation logic determined this event start precisely at 97.91 ms, ending at 100.86 ms, and having a duration of 2.95.Table 3 also shows automatic scoring for a 17.78 ms duration non-events immediately preceding and a 2.74 ms non-event immediately following the friction event. Table 3 is an example of a machine to machine data transfer which may be used to transfer automatically generated, selectively decimated, and meaningful information via wired or wireless means, such as from a digital transducer to a network or other host.

TABLE-US-00001 TABLE 1 Automatic scoring determining FIG. 8A event to be a friction type shear event. #segment 37.0000 38.0000 39.0000 Time begin [ms] 80.13 97.91 100.86 Auto Type NON EVENT FRICTION NON EVENT Duration [ms] 17.78 2.95 2.74 Total area 0.53 0.25 0.09 Time Skew 0.13 0.26 0.00 Big Spike 0.70 0.79 0.77 Ave Spike 0.64 0.60 0.77 Ramp Up 0.15 0.12 0.09 Ramp Down 0.53 0.25 0.09 Energy Skew 0.78 0.68 0.51 Friction factor A 4.03 8.09 3.54 Friction Factor B 3.60 4.98 4.49

Event Classification

[0080] Segmentation and analysis may be performed on any or all of the first rate of change data stream, the second rate of change data stream, and a third rate of change data stream. An event is may be subsequently characterized based on identified segments in the first rate of change data stream or second rate of change data stream, such as based on mathematic or visual characteristics of the segment. For example, as shown in FIG. 8, segments may be classified according to example event types. The event types shown in FIG. 8 are merely representative of some events, and it is understood that various other events may be represented either visually or mathematically using methods known in the art. Exemplary event types in FIG. 8 include I-dot, star, rocket, decay, multiple, peak, jerk, and friction. Similarly, as shown in FIG. 9, event types may be characterized according to data associated with the segments, such as decay, J-type, I type, and multiple events. The event types represented in FIGS. 8 and 9 may correspond to various phenomena on the object of interest such as negligible metal-to-metal contact, lubrication film failure, rolling contact, sliding contact, mixed mode lubrication regime, boundary lubrication regime, impact, severe sliding, and adhesion. FIGS. 10A-10D illustrate characterization of an event as a decay event. FIG. 10A is a plot of oversampled waveform data corresponding to an acceleration time waveform. FIG. 10B is a plot of values of a first rate of change data stream of the oversampled waveform data over a first interval and FIG. 10C is a plot of values of a second rate of change data stream of the oversampled waveform data over a second interval. FIG. 10D shows a plot of consecutive segments based on the second rate of change data stream of FIG. 10C. The plot of the segmented second rate of change data stream illustrates a shape and data corresponding to a decay-type event. The plot of FIG. 10D is a compressed data plot that may be subject to further characterization and analysis, as discussed in greater detail below.

Compressed Data Logic

[0081] A compressed data plot includes data related to segments of either or both of the first rate of change data stream and the second rate of change data stream. The compressed data plot includes data related to segments of the rate of change data streams in visual or numerical forms for displaying the compressed data plot on a display or for providing data related to the segmented rate of change data streams for further analysis on a processor. For example, the data shown in FIG. 9 may be transmitted from the sensor device 10 to a remote analysis device for further processing and characterization. Information related to the compressed data plot requires substantially less bandwidth to transmit from the sensor device 10 relative to the oversampled waveform data while providing detailed information related to a fast event detected in the oversampled waveform data.

[0082] In one embodiment, a threshold determination on the sensor device 10 defines a threshold to first designate whether a segment may correspond to a nonevent or an event exceeding a derived threshold that is subject to further analysis and reporting. FIG. 10D illustrates an exemplary threshold 400 having a value of approximately 1.0. The threshold may be defined and updated according to data of the first rate of change data stream or the second rate of change data stream. For example, the threshold may be defined by creating a data stream containing maximum rate of change data stream values over an interval by finding the maximum rate of change data stream value within a defined interval, such as 5 ms. The data stream is sorted from a maximum value to a minimum value. An early part of the sorted stream should represented maximum rate of change data stream values where no event is occurring. The threshold is further defined according to the method below.

[0083] Method A—determining where slope of sorted values increases dramatically [0084] Create a stream of slope values defined as (Sum of x values ahead −Sum of X values behind).sup.2, wherein x represents 30 ms of maximum values. [0085] Fit the first 50 ms of slope values to a straight line. [0086] Using the first 150 ms of sorted values, determine a standard deviation of a difference between the sorted stream of maximum values and a value of the straight line. [0087] A test value is defined as the determined standard deviation multiplied by a user-inputted factor. [0088] The stream of sorted maximum values is examined and a threshold value is defined as the maximum value whose slope exceeds the defined test value.

[0089] Method B—determining where sorted value departs from linear behavior [0090] Fit the first 100 ms of sorted values to a straight line. [0091] Using the first 80 ms of values to calculate standard deviation of a difference between the sorted values and the fit line. [0092] Test value defined as the standard deviation multiplied by the same user-inputted factor of Method A. [0093] The stream of sorted values is examined and the threshold value is defined as the first value whose difference from the fit line exceeds the test value.

[0094] Threshold value used [0095] If Method A produces a value, the Method A value is defined as the threshold value. [0096] If Method A does not produce a value but Method B does produce a value, the Method B value is defined as the threshold value. [0097] If neither Method A nor Method B produces a value, then a user-inputted threshold value is defined as the threshold value. [0098] A user-inputted threshold value may be defined as the threshold value to override the value of either Method A or Method B.

[0099] Embodiments of the present disclosure further provide for detecting and measuring effects of physiological phenomena including delineation of a threshold between normally distributed data and causal data. Evidence of causal data within a population is evident when a median is small compared with a mean value for the population. This occurs because phenomenological causal data often spans a much wider range with excursion to very high scores compared with Guassian normal data. Distributed data it typically found near the median. Causal data may be found near the median, near the mean, and at extreme. Extreme data is predominantly causal when causal data is present. Preferred embodiments of finding causal absolute value data include sorting a set of the data and evaluating a rate of change for the sorted distribution to select an inflection or other transition above which normally distributed data falls off and causal data increases.

[0100] Referring now to FIGS. 11A-11C, an example of determining a compressed data plot from an oversampled waveform data stream is shown. FIG. 11A shows a plot of oversampled waveform data 501 with a sampling rate of 102.4 kHz captured in a sensory measurement time waveform, in this example an acceleration time waveform. The oversampled waveform data is processed to derive a first rate of change data stream 500 and a second rate of change data stream 502, plots of which are shown in FIG. 11B. In the example of FIGS. 11A-11C, an interval of the first rate of change data stream 500 is approximately 500 μs, and an interval of the second rate of change data stream 502 is approximately 10 μs. FIG. 11C shows a compressed data plot of segments of the second rate of change data stream. The compressed data plot includes a threshold line 504 corresponding to a value of approximately 1.672, the threshold line designated according to methods described herein.

[0101] The compressed data plot of FIG. 11C shows illustrates segments corresponding to a phenomenon captured in the oversampled waveform data. For example, an event is represented between approximately 16 ms and 18 ms, wherein segments of the second rate of change data stream 502 show a friction-type event occurring within the oversampled waveform data. As shown in FIG. 11B, a value of the second rate of change data stream approaches a value of the first rate of change data stream, indicating an event characterized as a friction-type event.

[0102] FIG. 12 shows a data output generated on the sensor device 10 that corresponds to the compressed data plot of FIG. 11C. Output data includes various metadata related to the segments of the compressed data plot including a time that each segment begins, a duration of the segment, and other data describing each segment. Output data may also include an indication of the type of event within each segment, such as a friction event, jerk event, or other various events corresponding to characteristics of the segment. When a maximum value of the rate of change data stream within a segment does not exceed a designated threshold, the segment may be classified as a nonevent.

[0103] The compressed data plot of FIG. 11C and associated data shown in FIG. 12 compress the oversampled waveform data into segments and automatically classifies events based on the compressed data plot and related output data. The compressed data plot and associated output data significantly reduce an amount of data to be transmitted from the sensor device 10 relative to the oversampled waveform and many known methods of decimation.

[0104] In operation, the sensor device 10 receives oversampled waveform data from the sensor 12 in sensory contact with an object of interest, such as a machine. First and second rate of change data streams are derived with the processor 16 on the sensor device 10 from the received oversampled waveform data. One or both of the first and second rate of change data streams are analyzed by the processor 16 on the sensor device and segmented based on characteristics of the first and second rate of change data streams. Compressed data plot information is generated on the sensor device 10 including information related to each segment of the first and second rate of change data streams.

[0105] Referring to FIG. 13, the sensor device 10 preferably receives oversampled waveform data on a sensor device at step 602 and derives the first and second rate of change data streams at step 604 on the sensor device. Compressed data plot information is generated at step 606 on the sensor device based on one or both of the first and second rate of change data streams. At step 608 the compressed data plot information includes a determination of whether one or more segments of the compressed data plot are an event or a nonevent. The sensor device then transmits the compressed data plot information to a remote device for further analysis and characterization of the event at step 610.

[0106] Embodiments of the present disclosure may further include additional rate of change techniques that identify and characterize one or more fast events within oversampled waveform data. For example, a first rate of change data stream and second rate of change data stream may correspond to jerk values. A difference in a maximum amplitude and a minimum amplitude over a first interval and a second shorter interval are determined based on displacement or acceleration values measured on an object of interest. Values of the first rate of change data stream over the first interval may convey information about larger, slower motions such as powerful sliding, collision, and crushing contact. Values of the second rate of change data stream over the second shorter interval may convey information regarding very tiny and fast rubbing oscillations occurring.

[0107] Table 1 below is an example of data corresponding to five-step failure progression of a planar bearing and Table 2 is an example of data corresponding to five-stage failure progression of a roller bearing. Tables 2 and 3 report 160 ms peak-hold values measured as differences in a maximum amplitude and minimum amplitude of acceleration values within a first interval of 500 μs and a second shorter interval of 10 μs. The first rate of change data stream is referred to as Jerk500, while the second rate of change data stream is referred to as Jerk10. Differences in acceleration values are divided by a delta time between a point at which the maximum value occurs and the minimum value occurs. A peak hold duration is 0.16.

[0108] Data for planar bearing Jerk10 and Jerk500 are found to ramp up sharply after lubricant film boundary lubrication begins to fail between steps 1 and 2 during planar bearing steps. Data for roller bearing Jerk10 and Jerk500 demonstrate characteristic reductions during a brief period at stage 3 bearing failure progression.

TABLE-US-00002 TABLE 2 Planar Jerk10 (g/ms) Jerk500 (g/ms) Step 0 12,978 6 Step 1 19,831 6 Step 2 55,278 12 Step 3 125,690 86 Step 4 156,712 46

TABLE-US-00003 TABLE 3 Roller Jerk10 (g/ms) Jerk500 (g/ms) Stage 0 36,751 18 Stage 1 41,008 19 Stage 2 204,430 94 Stage 3 84,208 54 Stage 4 258,280 133

[0109] Embodiments of the present disclosure enable a sensor device that is in sensory communication with an object of interest to automatically monitor the object of interest and determine whether an event is occurring on the object of interest. If an event is detected by the sensor device, data may be reported by the sensor device to a remote device for further analysis. In one embodiment, the sensor device may locally process sensory data related to the object of interest and only transmit compressed data plot information when an event is detected, or otherwise generate an alert that an event has been detected on the object of interest.

[0110] The following non limiting examples illustrate the method and apparatus of the present disclosure identifying and characterizing events on an object of interest.

Example 1: Progression of a Planar Bearing Failure

[0111] Referring to FIGS. 14A-14E, a progression of a planar bearing failure is shown using methods of the present disclosure. In FIG. 14A, at an initial stage prior to substantial failure of the bearing is shown wherein both a first rate of change data stream 702 and a shorter second rate of change data 704 stream indicate low friction and steady values. A third rate of change data stream 706 having an interval between an interval of the first rate of change data stream 702 and the second rate of change data stream 704 also illustrates similar activity. A friction value 708, which is based on a relationship between the first rate of change data stream 702 and the second shorter rate of change data stream 704, is low and steady.

[0112] In FIG. 14B, at early stages of bearing failure the second shorter rate of change data stream 704 begins to show activity. The friction value 708 also begins to increase and has reduced steadiness. The third rate of change data stream 706 demonstrates increasing but small activity, and an amplitude of the first rate of change data stream 702 is increasing. Referring to

[0113] FIG. 14C, as failure progresses the second shorter rate of change data stream 704 shows increasing activity, with the friction value 708 having similar characteristic. The third rate of change data stream 706 shows increased activity, and the first rate of change data stream 702 increases in both amplitude and variability.

[0114] FIG. 14D illustrates continued progressive failure of the planar bearing, substantial fluctuations are present in the first rate of change data stream 702, second rate of change data stream 704, and third rate of change data stream 706. The friction value 708 also increases and includes wide fluctuation. Activity of the first rate of change data stream includes large amplitudes with decay and multiple event types.

[0115] Finally, FIG. 14E illustrates late stage failure of the planar bearing, with very high and defined events within the second shorter rate of change data stream 704 that correspond with the second rate of change data stream 704 and third rate of change data stream 706. Friction remains high, and the first rate of change data stream 702 includes extreme amplitudes with decay and multiple event types. Catastrophic failure is likely to occur quickly when the stage of FIG. 14E is reached.

Example 2: Progression of a Roller Bearing Failure

[0116] Referring to FIGS. 15A-15E, analysis of a progression of a planar bearing failure is shown using methods of the present disclosure. In FIG. 15A, a first rate of change data stream 802 includes some activity corresponding to mechanical vibrations but is otherwise within an acceptable range. A second rate of change data stream 804 is shown as being low and relatively steady. A third rate of change data stream 806 bears similarities to the second rate of change data stream 804. A friction value 808 is also low and steady.

[0117] In FIG. 15B, the second shorter rate of change data stream 804 begins to show activity as well as an increase in the friction value 808. The first rate of change data stream 802 and third rate of change data stream 806 include increase amplitude and variation.

[0118] Referring to FIG. 15C, the second shorter rate of change data stream 804 includes pronounced activity that mirrors the second rate of change data stream 806 and third rate of change data stream 808. Friction values 808 increase and include wide variation. The first rate of change data stream 802 includes high variation and amplitude showing well defined events.

[0119] In FIG. 15D, vibrations appear to subside for a brief interval. Friction values remain variable but lower than earlier stages, and the first rate of change data stream 802 events widen and become indistinguishable multiple event types. Subsequently, as shown in FIG. 15E, friction values 808 increase, and the first rate of change data stream 802 includes extreme amplitudes with decay and multiple event types. Catastrophic failure is likely to occur quickly when the stage of FIG. 15E is reached.

[0120] Reference is made herein to fast events or fast speed phenomena, intermediate events or intermediate speed phenomena, and slow events or slow speed phenomena. Embodiments of the present disclosure make use of oversampled waveform data to opportunistically or intentionally detect and measure fast, intermediate, and slow duration dynamic phenomena on an object of interest. Exemplary fast, intermediate, and slow events may preferably be defined in Table 4 as follows:

TABLE-US-00004 TABLE 4 Speed Fast phenomenon Intermediate phenomenon Slow phenomenon Duration range 0.04 to 1 ms 1 to 10 ms >10 ms Example Stress wave Stress wave Sustained input waves phenomena from critically from a not from a resonance or damped critically fluid-structure impact damped impact interaction or rubbing

[0121] Embodiments of the present disclosure include a method and apparatus for monitoring a variable or transient dynamic condition and includes a step of receiving digital sensory data that is physiologically responsive to a changing process, mechanism, operation, state, control, deterioration, health, strength, or other meaningful characteristic of interest. Digital measurement data is preferably oversampled at 102,400 samples per second or 204,800 samples per second, and includes data corresponding to displacement, acceleration, flux, voltage, or current.

[0122] Preferred embodiments of the present disclosure use oversampled waveform data to opportunistically or intentionally detect and measure fast, intermediate, and slow duration dynamic phenomena using a rate of change of short wavelength, mid wavelength, and long wavelength filters. Table 5 below shows preferred exemplary levels for stress waves, friction, impacts, and machine vibration events.

TABLE-US-00005 TABLE 5 Rate of change measurement interval Short Mid Long Interval duration 0.01 ms 0.05 ms 0.5 ms Typical sampling rate >100 samples/ms >100 samples/ms >100 samples/ms Example peak-to-peak 10.sup.3 to 5 g/s 10.sup.1-3 g/s 10.sup.−1 to 1 g/s (g/s) Example peak-to-peak (g) 10.sup.−2 to 0 g 10.sup.−2 to 1 g 10.sup.−1 to 1 g Nominal low pass <0.01 ms <0.05 ms <0.5 ms wavelength Nominal high pass >100 kHz >20 kHz >2 kHz frequency Steps to failure for sliding contact Rate of change Short Mid Long measurement interval Step 0 sliding failure Rate plot low and Rate plot mimics fast profile Rate plot normal for inactive more than slow operation (friction also low) Step 1 sliding contact Rate plot low and Rate plot has new activity Rate plot max active amplitudes increasing (friction increase) Step 2 sliding failure Rate plot new and Rate plot shows increasing Rate plot higher increased activity activity amplitudes with more (friction unsteady) variability Step 3 sliding failure Rate plot shows high Rate plot mimics slow profile Rate plot shows high events more than fast amplitudes with decay (high sporadic and multiple event friction) types Step 4 sliding failure Rate plot show very Rate plot mimics slow profile Rate plot shows high amplitudes with with multiple and decay event extreme amplitudes, defined events, some types decay, & multiple mimic intermediate event types—galling and low profiles etc. may be very soon Stages to failure for roller bearing Rate of change Short Mid Long measurement interval Stage 0 rolling contact Rate plot low and Rate plot mimics fast profile Rate plot normal for inactive more than slow operation (friction also low) Stage 1 rolling contact Rate plot low and Rate plot has new activity Rate plot max active similar to slow amplitudes increasing (friction increase) Stage 2 rolling contact Rate plot new and Rate plot follows slow profile Rate plot higher increased activity amplitudes with similar to defined event types intermediate and slow (friction up and swinging) Stage 3 rolling contact Rate plot gets better Rate plot mimics slow profile Rate plot levels drop, for a brief interval events widen and (friction is variable become less distinct and lower) Stage 4 Rate plot show very Rate plot mimics slow profile Rate plot shows rolling contact high amplitudes with with multiple and decay event extreme amplitudes, defined events, some types decay, & multiple mimic intermediate event types—galling and low profiles etc. may be very soon

[0123] FIG. 16 illustrates an exemplary plot with graph lines showing three low pass wavelength filters, which are also three high pass frequency filters, corresponding to 88 oversampled waveform data and associated activity of each filtered data plot. Also shown is a graph line estimating a friction, 80, indication derived from comparison among the other lines. In this example the lowest rate of change line, 82, is an envelope of low pass rate of change for wavelength 0.01 ms, the middle rate of change line, 84, is an envelope of low pass rate of change for wavelength 0.1 ms, and the uppermost rate of change line, 86, is an envelope of low pass rate of change for wavelength 0.5 ms. The area between abscissa and lowest line represents tiny displacement vibrations having vibration frequency content greater than approximately 100 kHz. The area between lowest and middle lines represents medium size displacements of moderate forcefulness having frequency content between approximately 10 kH and 100 kHz. The area between the middle and uppermost lines represents relatively large displacements having forceful effect and having frequency content between approximately 2 kHz and 10 kHz.

[0124] FIG. 17 illustrates a modified version of a Stribeck curve showing five regimes, including a logical extension for a not-lubricated or dry contact extension to the traditional four lubrication regime curve. Figurative elements include 90 five regimes of lubrication and contact; 91 coefficients for friction, wear, and life; 92 fluid film thickness , λ; 93 hydrodynamic (HD) and hydrostatic (HS) lubrication regime; 94 elastohydrodynamic (EHD) lubrication regime; 95 mixed mode lubrication regime; 96 boundary lubrication regime; 97 dry contact regime; 98 wear rate coefficient; and 99 surface remaining useful life coefficient. Dry, boundary & mixed may improve using thin film or solid film. Thin film may refer to ZDTP or other chemically deposited AW/EP oil additive. Thick film may refer to MoS.sub.2 or Graphite or PTFE or other solid substance used alone or as a liquid lubricant additive. Five regimes, not including component life and wear rate, was published in Society for Tribologists and Lubrication Engineers, “The Ultrasonic Tribologist—Using A Sonic/Ultrasonic Sensor To Monitor for Inadequate Lubrication,” 62nd Annual Meeting of the Society of Tribologists and Lubrication Engineers, 2007 pages 252-259. From right to left in FIG. 17 figuratively represents first and second lubrication regimes where a fluid film separation is typically present between moving surfaces while load bearing systems exhibit hydrodynamic (HD) and hydrostatic (HS) lubrication in first regime, and elestohydronamic (EHD) lubrication in the second regime. The fluid film typically becomes compromised in mixed and boundary lubrication regimes which causes increased friction, increased surface wear and decreased component life. Extreme contact in absence of even boundary lubrication regime rapidly destroys remaining useful life, rapidly increases wear rate, and drives friction to extreme such that surface movement leads to adhesion, severe sliding, galling and the like. Mixed mode, boundary, and dry contact are detectable using techniques described herein. It is most challenging to perceive sensory evidence of HS, HD, and EHD full fluid film separation; however thin film lubrication depending on extreme pressure (EP) and solid lubricants such as graphite, PTFE, and MoS.sub.2 may be detectable.

[0125] In one embodiment, methods of the present disclosure include logic including time of flight. Two or more sensor devices may receive a signal such as an event start. Stress waves originating from a source travel through material to the one or more sensor devices. Triangulation of signals is used to estimate time of flight from the source to each of the sensor devices. Material properties, such as speed of sound properties, may be used to further estimate distance to source from the one or more sensor devices. Time of flight triangulation may be performed on short, intermediate, and long event data.

[0126] Embodiments of the present disclosure may further include deductive logic comprising signal attenuation for each of short, mid, and long pass filtered wavelength rate of change data streams, as shown in Table 6 below:

TABLE-US-00006 TABLE 6 Rate of change Short Mid Long measurement interval Damping attenuation of Very high Medium Low stress wave by distance, layers, gaps, mount Attenuation of signal by Very high Indefinite Indefinite analyzer filters Attenuation through Very high Amplified rather than Indefinite tuned ultrasonic sensor attenuated Attenuation of ultrasonic Negligible Negligible/normal Negligible stress wave through common 3 to 8 kHz resonant piezoelectric accelerometer Attenuation of signal by Extreme Extreme Very high common MEMS accelerometer

[0127] In one embodiment, voting logic is a second, third, or more confirmation of a finding or evidence of a condition on an object of interest. Voting logic further confirms likelihood of a condition and reduces potential false positive or false negative results. False positive and false negative testing includes multiple measurements or techniques to arrive at a conclusion. Sensory voting logic may be used to confirm a positive or a negative finding using two or more different sensor devices. An independent technique using an exclusive or non-exclusive filtering of a single signal may be used to identify a potential false result. A false positive may be an incorrect determination that a finding is detected or alerted. A false negative may be an incorrect determination that a finding is not detected or alerted. An unwanted consequence of a false positive finding is the potential performance of unnecessary intrusive action. An unwanted consequence of a false negative finding may be the missing of an indication of a problematic condition. After alerting of one or more false positive or negative results, operators may ignore true positive or false indications. Oversampled data analysis provides plentiful measurement data such that a true result may be confirmed repeatedly to reduce a likelihood of false results.

[0128] The method of analyzing an oversampled data stream and sensor device of the present disclosure advantageously enables automatic monitoring of an object of interest, such as a machine, and provides reporting of information related to events on the object of interest. By analyzing oversampled data according to the method above, fast events of the oversampled data are analyzed to determine types of events on the machine. Embodiments of the present disclosure allow a sensor device to initially analyze an oversampled waveform and determine whether an event is occurring, and to subsequently output compressed data for further analysis. The sensor device may automatically classify an event type and report the event if the event is determined to exceed a desired threshold, thereby reducing the need of human monitoring of the device or performance of the object of interest. Embodiments of the present disclosure also reduce an amount of data transmitted from the sensor device. Embodiments of the present disclosure further enable analysis and classification of events on an object of interest corresponding to lubrication of components of a machine, as shown in FIG. 17.

[0129] Machinery health monitoring may incorporate a combination of vibration analysis and oil analysis techniques. Vibration analysis is typically measured within proximate vicinity to a measurement point. Vibration analysis is a dynamic measurement of characteristics such as rotor displacements in axial and radial movements (mm), velocity (mm/s), and acceleration (m/s.sup.2 or g). Frequency analysis is employed to monitor synchronous, nonsynchronous, and fluid film characteristics. Oil analysis is typically measured in proximity to a system of a lubricant flow path. Machinery health is monitored by sensing wear debris such as ferrous and nonferrous metals. System contamination is monitored for solid and liquid contaminants. Fluid health is monitored for changes to base oil and additives. Tribology is the study of surfaces in relative motion and of friction, lubrication, and wear. Neither oil analysis nor vibration analysis are focus on the tribological interface between surfaces in relative motion, friction, lubrication, and wear. Existing methods such as envelope vibration and PeakVue™ analyses of oversampled vibration data attempt to extend vibration reach to address point vicinity detection of periodic defects. However these techniques require manual setup, extra testing, special equipment, human input, and human interpretation; they are not automated. Passive ultrasonic detection is also used to detect nonspecific turbulence and friction within a proximate vicinity to the ultrasonic detector.

[0130] Preferred embodiments of the analysis technique of the present disclosure improve the state of the art and provide a bridge over this gap between oil analysis to deliver tribological information about surfaces in relative motion experiencing various degrees and types of friction, lubrication, and wear. An overarching theme for the present invention is a focus on surfaces in relative motion primarily using high frequency waveform analysis. These techniques are therefore point vicinity specific. They may be fully automated. Equipment information is not required. Human input is not required. Machine to machine transfer of data and compressed information is practical and expeditious. Surface defects may be revealed by type and subtype such. Information collected may be interpreted in categories such as shear, friction, impact, and layers to defect. Timestamp event data may be analyzed for periodicity and for random occurrences. Time of flight information can precisely locate distance and location of damage. Lubrication regime determination and dry contact evidence can be detected and used to recommend appropriate corrective actions.

[0131] The foregoing description of preferred embodiments of the present disclosure has been presented for purposes of illustration and description. The described preferred embodiments are not intended to be exhaustive or to limit the scope of the disclosure to the precise form(s) disclosed. Obvious modifications or variations are possible in light of the above teachings. The embodiments are chosen and described in an effort to provide the best illustrations of the principles of the disclosure and its practical application, and to thereby enable one of ordinary skill in the art to utilize the concepts revealed in the disclosure in various embodiments and with various modifications as are suited to the particular use contemplated. All such modifications and variations are within the scope of the disclosure as determined by the appended claims when interpreted in accordance with the breadth to which they are fairly, legally, and equitably entitled.