Virtual RPM sensor

• US 7,031,873 B2
• Filed: 05/16/2003
• Issued: 04/18/2006
• Est. Priority Date: 06/07/2002
• Status: Expired due to Fees

• Current Assignee: ExxonMobil Research & Engineering Company (Exxon Mobil Corporation)
• Original Assignee: ExxonMobil Research & Engineering Company (Exxon Mobil Corporation)
• Inventors: Song, Limin
• Primary Examiner(s): Barlow, John
• Assistant Examiner(s): LAU, TUNG S
• Application Number:
  US10/439,959
• Publication Number:
  US 20030229469A1
• Time in Patent Office: 1,068 Days
• Field of Search: 702/145, 702/189, 702/140, 702/150, 454/53, 454/51, 369/112, 369/47, 369/44, 369/43, 600/301, 600/300, 318/802, 318/561, 318/808, 318/807, 356/477, 700/279, 700/229, 736/63, 735/93, 738/61, 477/111, 385/12, 337/08, 434/51, 363/131, 348/208
• US Class Current: 702/145
• CPC Class Codes: G01P 3/44 for measuring angular speed G01P3/56 takes precedence

Abstract

The present invention is in a “virtual RPM sensor” which is preferably implemented by means of a software process. The software determines the RPM of a rotating machine using an adaptive signal processing algorithm. Preferably, the input to the software system of the present invention is digitized waveform data from either a vibration transducer on a rotating machine or vibration data stored in a database or file. The present invention permits the determination of the RPM of a rotating machine without the requirement of additional hardware so that readings may be obtained inexpensively, accurately and without interrupting machine operation.

Claims

1. A method for estimating the operating speed of a machine from time-based vibration data samples obtained from a vibration transducer, comprising the steps of:

(a) acquiring said time-based vibration data samples from said vibration transducer in physical constant with said machine;

(b) filtering said acquired time-based vibration data samples in order to obtain maximum low order vibration components while eliminating other components;

(c) estimating machine operating speed using said filtered time-based vibration data samples;

(d) fine tuning said machine operating speed estimate using a continuous vibration model; and

(e) providing said fine tuned machine operating speed estimate to a user.

2. The method of claim 1 wherein said fine tuned machine operating speed estimate is provided to an external device for condition monitoring.

3. The method of claim 1 wherein said time-based vibration data samples are obtained at a constant sampling rate.

4. The method of claim 1 wherein said time-based vibration data samples are obtained from a database containing sample data previously obtained from said machine.

5. The method of claim 1 wherein said filtering step is accomplished through the use of a digital filter.

6. A method for estimating the operating speed of a machine from time-based vibration data samples obtained from a vibration transducer, comprising:

(a) acquiring said time-based vibration data samples from said vibration transducer in physical constant with said machine;

(b) filtering said time-based vibration data samples in order to obtain maximum low order vibration components while eliminating other components;

(c) estimating machine operating speed using said filtered time-based vibration data samples, wherein said estimating operating machine speed is accomplished through the use of a digital comb filter;

(d) fine tuning said machine operating speed estimate using a continuous vibration model; and

(e) providing said fine tuned machine operating speed estimate to a user.

7. The method of claim 6 wherein said step of estimating machine speed further comprises the steps of:

(a) selecting a discrete range of operating speeds through which to conduct a search;

(b) dividing the time-based vibration data sample into a plurality of sub-records of sample length P;

(c) obtaining the coefficients of a digital comb filter based upon P;

(d) convoluting the time-based vibration data samples using the digital comb filter in the time-domain;

(e) computing the total vibration power and the coherent vibration power;

(f) computing a vibration power error function;

(g) selecting the optimal P that minimizes the error function by comparing the error functions of all sub-record lengths within said selected discrete range of operating speeds; and

(h) estimating the machine speed RPM based upon said selected optimal P.

8. The method of claim 7 wherein said total vibration power is determined by summing the squares of all sub-records and the coherent vibration power is determined by summing the squares of all time-averaged samples.

9. The method of claim 8 wherein the error function is determined by subtracting the coherent vibration power from the total vibration power.

10. The method of claim 7 wherein said machine speed is estimated as an RPM and is based upon said optimal P by dividing the sampling frequency by the optimal P value and multiplying the result by 60.

11. A method for estimating the operating speed of a machine from time-based vibration data samples obtained from a vibration transducer, comprising:

(a) acquiring said time-based vibration data samples from said vibration transducer in physical constant with said machine;

(b) filtering said time-based vibration data samples in order to obtain maximum low order vibration components while eliminating other components;

(c) estimating machine operating speed using said filtered time-based vibration data samples;

(d) fine tuning said machine operating speed estimate using a continuous vibration model; and

(e) providing said fine tuned machine operating speed estimate to a user,

wherein said fine tuning said machine operating speed further comprises;

(a) selecting a machine speed step length for searching RPM;

(b) selecting a search range of machine speed based oh the result of an optimal P obtained through the said estimating step;

(c) using a continuous vibration model of K'"'"'s vibration harmonics to represent the coherent vibration, in which each harmonic is a vibration component with a frequency as a multiple of RPM, and the number of harmonics are selected to include all possible vibration components in the filtered vibration samples to be processed;

(d) using a DFT to obtain the coefficients of said continuous vibration model from the filtered dine-based vibration data samples;

(e) computing the coherent vibration power from said continuous vibration model;

(f) computing the error function; and

(g) obtaining a machine operating speed measurement that minimizes the cost function.

12. The method of claim 11 wherein the cost function is the difference between overall vibration power and coherent vibration power.

13. The method of claim 11 wherein the cost function is determined by subtracting the coherent vibration power from the total vibration power.