Abstract
M.Ing. (Electrical And Electronic Engineering)
A brief introduction to laser Doppler anemometry is given. Signal
processing techniques for LOA are discussed with specific reference to the
salient advantages and disadvantages of each.
The abovementioned discussion motivates the use of adaptive filters as
parametric spectral estimators for frequency ext~action from Doppler
signals. Two adaptive algorithms ie. 'the least mean sqaures' (LMS) and
the 'recursive least sqaures' (RLS), are discussed. Results of a
real-time application of the LMS algorithm, as implemented on the
TMS320-10, as well as the results from a full study of the Kalman
algorithm are shown and discussed.
The implementation of a digital signal processor to make the application
of parametric spectral estimation to LOA viable is discussed under the
following headings: analogue module; mathematical processor;
communications module; storage an~ display unit.
The system is based qn a eight bit converter with conversion rates ranging
from the lower audio range to 20MHz. It has a versatile trigger system to
facilitate easy capture of Doppler data. Captured data is stored in an
onboard FIFO register and is available to the mathematical processing unit
for spectral analysis.
'/
The mathematical processing unit is a sixteen bit, multi-processor
environment specifically implemented for parametric spectral estimation.
The system comprises three TMS320-10 signal processors which are connected
by means of a high speed parallel communications channel. Processed data
is available to a display and storage system via a high speed interface
card that interfaces the sixteen bit signal processor environment to the
eight bit environment of the display device.