Complex signal decomposition and modeling

• US 8,239,170 B2
• Filed: 03/26/2008
• Issued: 08/07/2012
• Est. Priority Date: 03/09/2000
• Status: Expired due to Term

• Current Assignee: SmartSignal Corporation (Emerson Electric Company)
• Original Assignee: SmartSignal Corporation (Emerson Electric Company)
• Inventors: Wegerich, Stephan W.
• Primary Examiner(s): Dunn, Drew A
• Assistant Examiner(s): CHERRY, STEPHEN J
• Application Number:
  US12/055,471
• Publication Number:
  US 20080215291A1
• Time in Patent Office: 1,595 Days
• Field of Search: 702 70- 71, 702 73- 74, 702189-190
• US Class Current: 702/189
• CPC Class Codes: G06F 18/00 Pattern recognition
• CPC Class Codes: G06F 2218/12 Classification; Matching

Abstract

A system, method and program product for monitoring a complex signal for ultrasensitive detection of state changes, or for signature recognition and classification is provided. A complex signal is decomposed periodically for empirical modeling. Wavelet analysis, frequency band filtering or other methods may be used to decompose the complex signal into components. A library of signature data may be referenced for selection of a recognized signature in the decomposed complex signal. The recognized signature may indicate data being carried in the complex signal. Estimated signal data may be generated for determination of an operational state of a monitored process or machine using a statistical hypothesis test with reference to the decomposed input signal.

Claims

1. An apparatus for monitoring the operating condition of a system, comprising:

sensor means for acquiring a single time-varying signal characterizing operation of the system;

means for decomposing said single time-varying signal into a plurality of components represented by vectors of actual component values, each vector having actual component values each at least partially representing the same moment in time;

a memory for storing a plurality of reference snapshots of reference component values for known operating conditions; and

processor means responsive to the decomposing means, disposed to generate estimates of the component values using an example-based multivariate empirical modeling engine using a calculation that uses both the reference component values and the vectors of actual component values to calculate expected component values for each vector of actual component values, and

further disposed to generate residual values by differencing the actual component values and the expected component values, for determination of deviating operating conditions of the system.

2. An apparatus according to claim 1, wherein said decomposing means uses wavelet analysis to decompose the time-varying signal into coefficients.

3. An apparatus according to claim 1, wherein said decomposing means generates a power spectral density function for the time-varying signal to provide component signals representative of power levels related to a plurality of frequency ranges.

4. An apparatus according to claim 1, wherein said processor means further compares at least one said residual value to a threshold to determine deviating operating conditions of the system.

5. An apparatus according to claim 1, wherein said processor means further performs a sequential probability ratio test on successive values of at least one said residual value to determine deviating operating conditions of the system.

6. An apparatus according to claim 1, wherein said sensor means measures vibration of said system.

7. An apparatus according to claim 1, wherein said sensor means measures acoustic energy given off of said system.

8. An apparatus according to claim 1, wherein said sensor means measures electric current used in said system.

9. A method for monitoring the operating condition of a system, comprising the steps of:

acquiring a single time-varying signal characterizing operation of the system;

periodically extracting observations of features from the single time-varying signal, each feature relating to a component signal represented by actual component values, the actual component values each at least partially representing the same moment in time, and wherein the actual component values are wavelet coefficients for said single time-varying signal;

generating estimates, by a processor, for at least one of the extracted features responsive to extracting an observation of features, using an example-based multivariate empirical modeling engine using a calculation with both the actual component values and reference component values stored in a library of feature observations characteristic of acceptable operation of said system to calculate expected component values; and

differencing, by a processor, at least one of said expected component values with corresponding actual component values to produce residuals for determination of deviating operating conditions of said system.

10. A method according to claim 9, further comprising the step of determining whether at least one of said residuals exceeds a threshold.

11. A method according to claim 9, further comprising the step of performing a sequential probability ratio test on at least one of said residuals.

12. A method according to claim 9, wherein the step of extracting features includes generating a power spectral density function for said time-varying signal.