ESTIMATION OF UNCORRELATED CONTENT IN EXPERIMENTALLY MEASURED FREQUENCY-RESPONSE FUNCTIONS USING 3 MEASUREMENT CHANNELS

Experimental frequency response functions (FRF) measurements are normally made by measuring two signals: the force applied to the structure and the resulting response. If the coherence between these signals is less than unity, it is known only that there is uncorrelated content present in either or both signals. This paper presents, for the first time, a method for estimating the magnitude and source of any uncorrelated content present in a measurement system, provided that the original random signal is available for measurement. This is the case when a signal generator/amplifier/shaker set-up is used. However, this method cannot be applied to hammer impact testing. Measurements are made of the signals from the random noise generator, the force gauge, and the response transducer. From these measurements estimates can be made of uncorrelated content in the force and response measurements as well as in the amplification/shaker system. These estimates of uncorrelated content will be useful in improving the measurement system and in determining when the results from the 1 H ̂ and 2 H ̂ estimators are valid. They also can be used to estimate the s2 term in s H ̂. The uncorrelated content estimators are applied in several different FRF measurements. Estimates of uncorrelated content from an inefficient non-linear shaker-to-force-gauge connection, from a ground-loop problem, and from leakage are shown. Limits on dynamic range are discussed. © 1990.