This chapter describes some common operations performed on measured data that are either part of the system identification process or preparatory steps. Most of the techniques presented here have very general applicability, but were selected because they have been found useful for tasks related to aircraft system identification. The selected techniques certainly do not represent the full spectrum of techniques available. Ref. 1 is an excellent resource for techniques of the type described here.

Sections 11.1 and 11.2 discuss filtering and smoothing, which are concerned with separating deterministic signal from random noise for measured time series. This is an important operation in system identification, because the general aim is to identify a mathematical model based on the deterministic parts of the measurements. Section 11.3 explains the link between the smoothing methods and interpolation, which provides a means to reconstruct missing or bad data points. Specialized methods for smoothed numerical differentiation are described next. These methods find use in equation-error modeling and sensor position error correction. Practical methods for computing the finite Fourier transform and power spectral estimates, which form the basis for the frequency-domain methods of Chapter 7, are also discussed in detail. Finally, the chapter concludes with a discussion of methods for comparing signal waveforms and visualizing aircraft motion during a flight-test maneuver using animated computer graphics.