Method and system for removing very low frequency noise from a time-based data set

• US 7,321,845 B2
• Filed: 04/13/2005
• Issued: 01/22/2008
• Est. Priority Date: 01/24/2003
• Status: Expired due to Term

• Current Assignee: Pratt & Whitney Canada Corporation (Rtx Corporation)
• Original Assignee: Pratt & Whitney Canada Corporation (Rtx Corporation)
• Inventors: Collin, Denis, Matschek, Peter, Nguyen, Phuc Luong
• Primary Examiner(s): Ramos-Feliciano; Eliseo
• Assistant Examiner(s): SUAREZ, FELIX E
• Application Number:
  US11/104,645
• Publication Number:
  US 20050209821A1
• Time in Patent Office: 1,014 Days
• Field of Search: 702/61, 702/104, 702/122, 702/179, 702181-183, 702/186, 702/195, 435/7.1, 703/5, 709/217, 375/136
• US Class Current: 702/183
• CPC Class Codes: G05B 23/0221 Preprocessing measurements, e.g. data collection rate adjustment; Standardization of measurements; Time series or signal analysis, e.g. frequency analysis or wavelets; Trustworthiness of measurements; Indexes therefor; Measurements using easily measured parameters to estimate parameters difficult to measure; Virtual sensor creation; De-noising; Sensor fusion; Unconventional preprocessing inherently present in specific fault detection methods like PCA-based methods
• CPC Class Codes: G06F 2218/02 Preprocessing
• CPC Class Codes: G06F 2218/12 Classification; Matching

Abstract

A method and system are disclosed for removing very low frequency noise from a time-based data set for use in data analysis of the time-based data set having a plurality of data points, the method comprising providing the time-based data set to be analyzed, determining a baseline value from at least a portion of the data set, smoothing the data set to diminish an effect of extraneous data points and obtain smoothed data, wherein the smoothing includes using a Fast Fourier Transform (FFT) algorithm to transfer the time-based data set into a frequency based data set, attenuating low frequencies of the frequency-based data set and using an inverse Fourier Transform (FFT) algorithm to transfer the attenuated frequency based data set into an attenuated time-based data set.

Claims

1. A method for removing very low frequency noise from a time-based data set for use in data analysis of said time-based data set having a plurality of data points, comprising:

providing said time-based data set to be analyzed;

determining a baseline value from at least a portion of said data set;

smoothing the data set to diminish an effect of extraneous data points and obtain smoothed data, wherein said smoothing includes;

using a Fast Fourier Transform (FFT) algorithm to transfer said time-based data set into a frequency based data set;

attenuating low frequencies of said frequency-based data set; and

using an inverse Fourier Transform (IFT) algorithm to transfer said attenuated frequency based data set into an attenuated time-based data set; and

displaying the smoothed data using a display unit.

2. A method as claimed in claim 1, wherein said attenuating comprises selling at least one lowest FFT coefficient to zero, thereby negating an effect of said low frequency.

3. A method as claimed in claim 2, wherein said at least one lowest FFT coefficient is two lowest FFT coefficient.

4. A method as claimed in claim 1, wherein said low frequencies represent a seasonal effect.

5. A method as claimed in claim 1, further comprising, after said step of providing said data set:

providing predicted values corresponding to said data set;

calculating a difference between each of said data points of said data set and said predicted values to obtain difference data;

wherein said step smoothing is carried out using said difference data.

6. The method as claimed in claim 1, further comprising:

comparing said smoothed data to said baseline data to detect a change;

evaluating said change to determine an occurrence of a given condition, and

generating an alert message corresponding to the given condition if the occurrence of the given condition is determined.

7. A system for removing very low frequency noise from a time-based data set for use in data analysis of said time-based data set having a plurality of data points, comprising:

a data provider for providing said time-based data set to be analyzed;

a baseline determiner for determining a baseline value from at least a portion of said data set;

a low frequency noise data smoother for smoothing the data set to diminish an effect of extraneous data points and obtain smoothed data, wherein said smoothing includes;

using a Fast Fourier Transform (FFT) algorithm to transfer said time-based data set into a frequency based data set;

attenuating low frequencies of said frequency-based data set; and

using an inverse Fourier Transform (IFT) algorithm to transfer said attenuated frequency based data set into an attenuated time-based data set; and

a display unit for displaying said smoothed data.

8. A system as claimed in claim 7, wherein said attenuating comprises setting at least one lowest FFT coefficient to zero, thereby negating an effect of said low frequency.

9. A system as claimed in claim 8, wherein said at least one lowest FFT coefficient is two lowest FFT coefficient.

10. A system as claimed in claim 7, wherein said low frequencies represent a seasonal effect.

11. A system as claimed in claim 7, further comprising:

a comparison provider for providing predicted values corresponding to said data set;

a difference calculator for calculating a difference between each of said data points of said data set and said predicted values to obtain difference data, wherein said difference data is provided to said odd-man out locator for said locating.

12. The system as claimed in claim 7, further comprising:

a change determiner to detect a change by comparing said smoothed data to said baseline data;

a problem identifier to determine an occurrence of a given condition by evaluating said change and;

an alert generator for generating an alert message corresponding to the given condition if the occurrence of the given condition is determined.