Stability Spectral Analysis Based on the Damping Spectral Analysis and the Data from Dryden flight tests, ATW_f5_m83h10-1
Location of the Test Wing
Details of the test wing
Test Video
The Full Data : atw_f5_m83h10_1 Details
IMF Data x83 : atw_f5_m83h10_1
Fourier Spectra of Various Sum of IMFs
Hilbert Filtered Data x83 : atw_f5_m83h10_1
Hilbert Spectrum : x83
Hilbert Spectrum : x83 Details
Spectrogram (512) : x83
Spectrogram (512) : x83 Details
Spectrogram (1024) : y83
Spectrogram (1024) : y83 Details
Re-sampled Hilbert Filtered Data : y(I)
Mean Hilbert Spectrum : y(i)
Hilbert Spectrum : x83 Details
Marginal Hilbert and Fourier Spectra : y83
3D Mean Hilbert Spectrum : y(i)
3D Spectrogram : y83
Instantaneous frequency : y(i)
Instantaneous frequency : y(i) Details
Mean Hilbert and Spectrogram : y83
Mean Hilbert and Spectrogram : y83 Details
Envelopes of Data x83 and Filtered Data y83
Instantaneous frequency and data Envelope
Stability Spectrum Problems of the previous approach: –1. Hilbert Envelope contains modulation in the amplitude –2. Define both positive and negative damping –3. How to define the instantaneous frequency
Time-Frequency Dependent Damping Analytic function of k th mode : (subscripts omitted for simplicity) Model time-dependent decay factor: Loss factor: where (t) is critical damping ratio and 0 (t) is natural frequency is the (damped) system frequency If = const., = 1/2 t --- Under damped harmonic oscillator
Hilbert Damping Spectrum Time and Frequency Dependent Damping - (t) contoured on the time- frequency plane, i.e. [ (t), (t), t] ( , t), where (t)= ( (t), t), Time-averaged loss factor using root-mean-square: A frequency dependent damping loss factor can be calculated if the system is essentially linear: If n is a resonant frequency of a structure, ( n ) = loss factor obtained using conventional modal method
Difference in Envelopes Hilbert Transform vs Spline
Different Envelopes
Different Derivatives from Envelopes
Different Derivatives from Envelopes : S5
Different Derivatives from Envelopes : S21
Stability Spectrum [n, t, f]=isspec(imf(:, 1:7), 600, 0, 30, 0, tt(11750), 20, 0.01,[],'no','no'); Data from C1; Frequency resolution : 600 Frequency range : 0 to 30 Hz Smoothed temporally with 20 point = 0.2 Second In this study NT= 3,5,10,15,20 were used Cut-off magnitude set 0.01 In this study PER=0.1, 0.01, 0.005, were used
Effect of Magnitude Cut-off Varying percentage cut-off values
Hilbert Stability Spectrum : Per=0.001, NT=10
Hilbert Stability Spectrum : Per=0.005, NT=10
Hilbert Stability Spectrum : Per=0.01, NT=10
Hilbert Stability Spectrum : Per=0.1, NT=10
Stability Index as a Function of Frequency Per=0.1, 0.01, 0.005, 0.001
Stability Index as a Function of Time Per=0.1, 0.01, 0.005, 0.001
Effect of Smoothing Varying NT values
Hilbert Stability Spectrum : Per=0.01, NT=3
Hilbert Stability Spectrum : Per=0.01, NT=5
Hilbert Stability Spectrum : Per=0.01, NT=10
Hilbert Stability Spectrum : Per=0.01, NT=15
Hilbert Stability Spectrum : Per=0.01, NT=20
Hilbert Stability Spectrum : Per=0.01, NT=30
Stability Index as a Function of Frequency NT = 3, 5, 10, 15, 20
Stability Index as a Function of Time NT = 3, 5, 10, 15, 20
Nonlinearity Determined from various methods: HHT Teager’s Energy Operator Generalized Zero-crossing
IF from Various Methods
IF from Various Methods, More Details
Preliminary Conclusions The flutter is quite nonlinear. The flutter frequency increases with increasing Mach number. Even from Fourier point view, there is a faint sub-harmonics vibration for the flutter, which usually suggests nonlinearity. Nonlinearity becomes obvious toward the end of the test, after the flutter amplitude increases almost exponentially and starts to level off. The nonlinear vibration is confirmed by Fourier based spectrogram, which clearly shows second harmonics. Just before the shattering of the wing, the flutter frequency starts to decrease suggesting yielding of the wing. The frequency change at the end cannot be detected quantitatively by any method other than Hilbert Spectral Analysis. Stability spectra with different magnitude cut-off and smoothing: tentative guide: PER>0.01; NT<20. Over most of the range, the wing is unstable with negative stability index (i.e. negative damping).