Iterative Deconvolution of Acoustic Source Maps - Accuracy, Speed and Extension to the Third dimension
Marthijn Tuinstra is a PhD student in the Department of Engineering Fluid Dynamics. (Co)Promotors are prof.dr.ir. C.H. Venner and dr.ir. M.P.J. Sanders from the Faculty of Engineering Technology and dr. H. Brouwer from NLR.
The presented research aims to advance iterative deconvolution methods for aeronautical testing in wind tunnels, a crucial tool for the evaluation of aeroacoustic performance of new aircraft concepts or components. The investigation focuses on acoustic imaging techniques and addresses signal distortion, deconvolution methods, and source location algorithms.
Key findings are:
- A comparison between the exact solution and the approximate ray solution for sound transmission through shear layers, reveals that uncertainties are negligible in most common situations.
- A random phase screen approach was used to model signal distortions in relation to time delay statistics, furthermore establishing a relation between scattering frequency and wave orientation. This approach was validated through wind tunnel noise measurements, demonstrating high accuracy.
- The acoustic deconvolution problem was analysed, and splitting methods were generally found to be robust, neither amplifying nor attenuating the singular components. However, accidentally introduced high spatial frequency singular components are retained in the converged solution.
- The simultaneous JOR and randomised ART relaxation schemes are most robust, resulting in minimal distortion and introduction of singular components in the solution .
- A highly efficient deconvolution algorithm, MAID, was developed, that employs acceleration methods based on multilevel matrix multiplication, solution of an equivalent problem, and domain truncation.
- Using the principles of computed tomography, a three-dimensional acoustic source location algorithm, ALTRE, was developed, which was successfully demonstrated through validation with both synthetic and experimental data sets.
In conclusion, significant contributions were made in the assessment of aeroacoustic measurement uncertainties and the developed deconvolution algorithms. MAID and ALTRE, offer improved capabilities for noise assessment and source location in wind tunnels.
