1. Vibrationdata 1 Unit 5 The Fourier Transform

2. Vibrationdata 2 Courtesy of Professor Alan M. Nathan, University of Illinois at Urbana-Champaign

3. Vibrationdata 3 Unit 5 Many, many different equations for Fourier Transforms..... 1. Mathematician Approach 2. Engineering Approach Even within each approach there are many equations.

4. Vibrationdata 4 Continuum of Signals n Random Stationary vibration signals can be placed along a continuum in terms of the their qualitative characteristics n A pure sine oscillation is at one end of the continuum n A form of broadband random vibration called white noise is at the other end n Reasonable examples of each extreme occur in the physical world. n Most signals, however, are somewhere in the middle of the continuum

5. Vibrationdata 5 Frequency Example

6. Vibrationdata 6 Resolving Spectral Components n The time history appears to be the sum of several sine functions n What are the frequencies and amplitudes of the components? n Resolving this question is the goal of this Unit

7. Vibrationdata 7 Equation of Previous Signal n At the risk of short-circuiting the process, the equation of the signal in the previous figure is n The signal thus consists of three components with frequencies of 10, 16, and 22 Hz, respectively. n The respective amplitudes are 1.0, 1.5, and 1.2 G n In addition, each component could have had a phase angle n In this example, the phase angle was zero for each component n Need some sort of "spectral function" to display the frequency and amplitude data n Ideally, the spectral function would have the form shown in the next figure

8. Vibrationdata 8

9. Vibrationdata 9 Textbook Fourier Transform The Fourier transform X(f) for a continuous time series x(t) is defined as where -  < f <  The Fourier transform is continuous over an infinite frequency range. Note that X(f) has dimensions of [amplitude-time].

10. Vibrationdata 10 Textbook Inverse Fourier Transform The inverse transform is Thus, a time history can be calculated from a Fourier transform and vice versa.

11. Vibrationdata 11 Magnitude and Phase Note that X(f) is a complex function. It may be represented in terms of real and imaginary components, or in terms of magnitude and phase. The conversion to magnitude and phase is made as follows for a complex variable V.

12. Vibrationdata 12 Real Time History n Practical time histories are real n Note that the inverse Fourier transform calculates the original time history in a complex form n The inverse Fourier transform will be entirely real if the original time history was real, however.

13. Vibrationdata 13 Fourier Transform of a Sine Function The transform of a sine function is purely imaginary. The real component, which is zero, is not plotted. f = 0  Dirac Delta function

14. Vibrationdata 14 Characteristics of the Continuous Fourier Transform n The real Fourier transform is symmetric about the f = 0 line n The imaginary Fourier transform is antisymmetric about the f = 0 line.

15. Vibrationdata 15 Characteristics of the Continuous Fourier Transform n This Unit so far has presented Fourier transforms from a mathematics point of view n A more practical approach is needed for engineers ………

16. Vibrationdata 16 Discrete Fourier Transform n An accelerometer returns an analog signal n The analog signal could be displayed in a continuous form on a traditional oscilloscope n Current practice, however, is to digitize the signal, which allows for post-processing on a digital computer n Thus, the Fourier transform equation must be modified to accommodate digital data n This is essentially the dividing line between mathematicians and engineers in regard Fourier transformation methodology

17. Vibrationdata 17 Discrete Fourier Transform Equations The Fourier transform is The corresponding inverse transform is

18. Vibrationdata 18 Discrete Fourier Transform Relations Note that the frequency increment  f is equal to the time domain period T as follows The frequency is obtained from the index parameter k as follows where k is the frequency domain index

19. Vibrationdata 19 Nyquist Frequency n The Nyquist frequency is equal to one-half of the sampling rate n Shannon’s sampling theorem states that a sampled time signal must not contain components at frequencies above the Nyquist frequency n Otherwise an aliasing error occurs

20. Vibrationdata 20 Half Amplitude Discrete This is the imaginary component of the transform. The real component is zero.

21. Vibrationdata 21 Previous Fourier Transform Plot n Note that the sine wave has a frequency of 1 Hz n The total number of cycles is 512, with a resulting period of 512 seconds n Again, the Fourier transform of a sine wave is imaginary and antisymmetric. n The real component, which is zero, is not plotted.

22. Vibrationdata 22 Spectrum Analyzer Approach n Spectrum analyzer devices typically represent the Fourier transform in terms of magnitude and phase rather than real and imaginary components n Furthermore, spectrum analyzers typically only show one-half the total frequency band due to the symmetry relationship n The spectrum analyzer amplitude may either represent the half-amplitude or the full-amplitude of the spectral components n Care must be taken to understand the particular convention of the spectrum analyzer n Note that the half-amplitude convention has been represented in the equations thus far

23. Vibrationdata 23 Full Amplitude Fourier Transform The full-amplitude Fourier transform magnitude would be calculated as

24. Vibrationdata 24 Full Amplitude Fourier Transform n Note that k = 0 is a special case n The Fourier transform at this frequency is already at full-amplitude n Consider a sine wave with an amplitude of 1 G and a frequency of 1 Hz n This signal would simply have a full-amplitude Fourier magnitude of 1 G at 1 Hz, as shown in the next figure

25. Vibrationdata 25

26. Vibrationdata 26 Full Amplitude Fourier Transform Time History Example

27. Vibrationdata 27 Full Amplitude One-Sided

28. Vibrationdata 28 Unit 5 Exercise 1 A time history has a duration of 20 seconds. What is the frequency resolution of the Fourier transform?

29. Vibrationdata 29 Unit 5 Exercise 2 a) Generate sine time history using: Amplitude = 7, Frequency = 2 Hz, Duration 100 sec, Sample rate = 100 b) Save sine function as ASCII text file. c) Then take the Fourier transform of sine function: vibrationdata > Fourier transform mean removal=yes, window=rectangular, min freq = 0 Hz, max freq = 10 Hz

30. Vibrationdata 30 Unit 5 Exercise 3 Calculate Fourier transforms and identify dominant frequencies for these time histories from previous webinar units: tuning_fork.txt - time (sec) & unscaled sound pressure drop.txt - time (sec) & accel(G) channel.txt - time (sec) & accel(G) Use: vibrationdata > Fourier transform mean removal=yes, window=rectangular, Min Freq=0 Hz DataMax Freq (Hz) tuning_fork1000 drop100 channel1000