Wavelet-Processed Flight Data for Robust Aeroservoelastic Stability Margins

Abstract

Wavelet analysisfor ltering and system identi cation is used to improve the estimation of aeroservoelastic (ASE) stability margins.Computationof robust stability margins for stability boundaryprediction depends on uncertainty descriptions derived from the test data for model validation. Nonideal test conditions, data acquisition errors, and signal processing algorithms cause uncertainty descriptions to be intrinsically conservative. The conservatism of the robust stability margins is reduced with parametric and nonparametric time-frequency analysis of ight data in the model validation process. Nonparametric wavelet processing of data is used to reduce the effects of external disturbances and unmodeled dynamics. Parametric estimates of modal stability are also extracted using the wavelet transform. F-18 High Alpha Research Vehicle ASE ight test data are used to demonstrate improved robust stability prediction by extension of the stability boundary from within the ight envelope to conditions suf cently beyond the actual ight regime. Stability within the ight envelope is con rmed by ight test. Practical aspects and guidelines for ef ciency of these procedures are presented for on-line implementation. Nomenclature a, a i = wavelet scale, indexed scale values (dimensionless) F.P; 1/ = feedback interconnection structure g = wavelet basis function K = feedback control system P.s/; O P.s/ = Laplace transform of system plant, estimate W add ; W in ; W ns = weightings on 1 add , 1 in , noise W g = continuous wavelet transform with basis g X .¿; !/; O X.¿; !/ = wavelet-transformed signal, ltered signal X .!/; O X .!/ = frequency-domain signal, estimate x.t /; O x.t / = time-domain signal, ltered signal 0 = robust stability margin 1; O 1 = uncertainty operator, estimate ± N q = uncertainty in ight condition = damping ratio ¹ = structured singular value ¿ = wavelet translation time Á.t /; Á 0 = signal phase, constant phase lag ! d ; ! n = modal damped and natural frequency ! 0 = wavelet peak frequency

Match-Point Solutions for Robust Flutter Analysis

“…7 Another application is the analysis of aeroservoelasticity for ight vehicles. 8 This paper considers the analysis of utter; however, the match-point formulation may be directly used for all of the applications of ¹-method analysis.…”

Robust flutter analysis of a nonlinear aeroelastic system with parametric uncertainties

“…A real flight test of the aerostructure test wing was performed to evaluate the flutterometer and the results demonstrated the validity and safety of this approach [21]. To reduce the conservatism of the results generated by robust flutter margin prediction, wavelet analysis and Volterra modeling approach have been developed and more credible predictions were obtained [9,19,22]. In literature [15], a match point solution has been achieved by setting airspeed as the perturbation variable in stead of dynamic pressure such that the flight condition of the model depends on a single parameter.…”

Non-Stationary Dynamics Data Analysis With Wavelet-SVD Filtering

“…Multisine waveforms have a chirp-like quality, but with a burst of high-frequency chirp at the end of the signal [3]. Figure 20 (top plot) shows the nature of a multisine in the accelerometer response for a 35-to-5 Hz sweep.…”

“…In this application, the time-frequency distribution, Qðt; f Þ; is constructed from a multiscale wavelet transform [3]. Therefore, the distribution is not the strict form of a joint energy density since the requirement on marginals is not enforced.…”

“…Transfer functions and modal parameter estimates derived from time-frequency representations have been applied to estimate state-space aeroservoelastic models [2][3][4]. Time-frequency structure is revealed by quantifying the distribution of signal energy as a joint function of time and frequency.…”