Localization and identification of moving noise sources

Sound source localisation methods are widely used in the automotive, railway, and aircraft industries. Many different methods are available for the analysis of sound sources at rest. However, methods for the analysis of moving sound sources still suffer from the complexities introduced by the Doppler frequency shift, the relatively short measuring times, and propagation effects in the atmosphere. The project LION combines the expertise of four research groups from three countries working in the field of sound source localisation: The Beuth Hochschule für Technik Berlin (Beuth), the Turbomachinery and Thermoacoustics chair at TU-Berlin (TUB), the Acoustic Research Institute (ARI) of the Austrian Academy of Sciences in Vienna and the Swiss laboratory for Acoustics / Noise Control of EMPA. The mentioned institutions cooperate to improve and extend the existing methods for the analysis of moving sound sources. They want to increase the dynamic range, the spatial, and the frequency resolution of the methods and apply them to complex problems like the analysis of tonal sources with strong directivities or coherent and spatially distributed sound sources. The partners want to jointly develop and validate these methods, exploiting the synergy effects that arise from such a partnership. Beuth plans to extend the equivalent source method in frequency domain to moving sources located in a halfspace, taking into account the influence of the ground and sound propagation through an inhomogeneous atmosphere. ARI contributes acoustic holography, principal component analysis, and independent component analysis methods and wants to use its experience with pass-by measurements for trains to improve numerical boundary element methods including the transformation from fixed to moving coordinates. TUB develops optimization methods and model based approaches for moving sound sources and will contribute its data base of fly-over measurements with large microphone arrays as test cases. EMPA contributes a sound propagation model based on Time-Variant Digital Filters with particular consideration of turbulence and ground effects and will also generate synthetic test cases for the validation of sound source localization algorithms. The project is planned for a period of three years. The work program is organized in four work packages: 1) the development of algorithms and methods, 2) the development of a virtual test environment for the methods, 3) the simulation of virtual test cases, and 4) the application of the new methods to existing test cases of microphone array measurements of trains and aircraft.

To improve the calcution of the acoustic radiation of objects, it is important to know the source positions with a higher resolution. In the transportation sector we deal with moving sources. Recording the crossing of a moving obect with a stationary acoustic camera, we have to take the Doppler effect into acount. The Doppler effect gives a smeared response in the frequency domain using long time recordings from the microphones of the acoustic camera. Using numerical methods (2.5D boundary element method) the transfer paths from the source positions to the microphone of the acoustic camera are calculated. In the frequency domain the signals are projected to the assumed source positions using the calculated transfer functions. The Doppler effect and the side effects of the nuermical transformation of the signals from the time to the freuqency domain are taken into account in the calculated transfer functions. The smeared responses allow to use several frequencies in the vincinity of the original frequency. This method allows to increase rthe resolution by far. The resolution in the direction of movement is even higher than the resolution perpendicular to it.