Separation of Sine and Random Components from Vibration Measurements

Abstract

Defining sine and random environments for test or analysis requires measurement on a test article for which these components of the environments are superimposed with electrical noise and structural modal responses. Examples of such combined environments are rocket motors, helicopters, and propeller airplanes. Sine and random specifications have typically been defined by enveloping spectra from the combined environment, which may lead to overly-conservative specifications and possible overtesting and overly-conservative analyses. This paper presents a signal processing method that accurately separates the multiple sinusoidal components and their harmonics from the background random. This method is validated and demonstrated using signals with known sine and random components both as clean data and with the superposition of electrical noise and structural dynamic modal response. Speed profiles for sources of the sinusoidal excitations are used to identify the sinusoidal excitations and the Vold-Kalman filter is used to extract order time histories. The extracted sine and random spectra may be used to define accurate environments for testing or analysis. Component finite element analyses in particular are usually performed separately for random and sine. More accurate environments will lead to more accurate margins and more efficient designs.