FEATURES OF A COMBINED FFT AND HILBERT TRANSFORM FOR PHASE DOPPLER SIGNAL-PROCESSING

Citation
P. Lehmann et Eh. Schombacher, FEATURES OF A COMBINED FFT AND HILBERT TRANSFORM FOR PHASE DOPPLER SIGNAL-PROCESSING, Measurement science & technology, 8(4), 1997, pp. 409-421
Citations number
23
Categorie Soggetti
Instument & Instrumentation",Engineering
ISSN journal
09570233
Volume
8
Issue
4
Year of publication
1997
Pages
409 - 421
Database
ISI
SICI code
0957-0233(1997)8:4<409:FOACFA>2.0.ZU;2-S
Abstract
Phase Doppler anemometry (PDA) is a laser optical method to determine particle diameters and velocities within two-phase or multiphase flows simultaneously and with high spatial and temporal resolution. The mea sured phase difference between two Doppler bursts is related to the pa rticle diameter. The proposed method for signal processing is based on a combined Hilbert transform and Fast Fourier transform (FFT) phase D oppler burst analysis. The numerical determination of the burst envelo pes gives an estimation of the time delay between two related signals or, more generally, of the burst maximum position within the record ti me. This estimate is completed by the conventional FFT based signal an alysis which is used to estimate frequency and phase difference. By th is two-step estimation the restriction to the [0, 360 degrees] interva l resulting from conventional signal processing can be avoided. The fe asibility of the method in terms of an on-line determination of absolu te phase differences is investigated by Monte Carlo simulations and de monstrated by means of a standard PDA arrangement with ball lenses rot ating as spherical particles through the measurement volume. Phase dif ferences up to 2500 degrees were determined reliably. In addition, alg orithms to calculate Doppler frequencies and Doppler intensities, burs t lengths and further characteristic parameters taking the signal qual ity into account are introduced and studied by Monte Carlo simulations . Based on these results validation strategies to reduce the influence of maltriggered bursts, trajectory or Gaussian beam effects on measur ement results can be developed. Furthermore, by using the information of burst length and maximum position, optimized signal processing algo rithms can be realized in order to achieve maximum accuracy in frequen cy, phase difference and signal-to-noise ratio (SNR) estimation.