Error Analysis of Cash Method:Computer-Averaging of Noisy Periodic Signals.

摘要

A computer-assisted synchronous hot wire, or CASH, method was developed to permit computer averaging of hot wire data when sampling oscillating velocity profiles. Using emitted sound pressure as phase reference, the computer separates the periodic component of a hot wire signal from the turbulent background by averaging over a large number of acoustic cycles. So far, the technique has been utilized in two investigations: (1) interface oscillations of flow-excited cavities, and (2) growth of instability waves in laminar jets. Accuracy depends on the ability to match closely the reference source frequency. Any mismatch produces systematic error. For a given amount of mismatch, however, error is minimized by dividing the sample into subsets and performing the averaging at two levels. In the cavity resonator problem, random deviations from periodicity in the reference itself make frequency-matching difficult. Measured standard deviations show a predicted periodicity, with greatest deviation occurring 90 deg after signal maximum. By counting beats against a standard oscillator, frequency error can be estimated at data acquisition time. With this information, sync error in average signal can be completely removed by post-run data reduction. (Author)