Acoustic noise source modelling based on microphone array measurements

A method will be presented which allows a construction of an acoustic sourc e model based on the analysis of microphone array measurements during a tra in pass-by. The conventional array beam-forming technique is used as a kind of pre-processing or first order analysis to calculate in a second step th e true source strengths by a back-projection method. The final equation to be solved connects the source power distribution on the vehicle surface wit h the measured microphone array amplitudes via a Green's function transfer matrix. By these means the side-lobe effects of the array characteristics a s well as the Doppler frequency shift and amplitude augmentation of the sou rces are removed. The model is updated for consecutive time slices during t he pass-by so that it is possible to study directional and time-dependent e ffects on the source characteristics. The method has been applied to measur ements on the ICE-V train where three known artificial noise sources (louds peakers) had been mounted on the outside of the sidewall. The source power distribution function gives a much better source separation than the conven tional beam-forming result. The calculated source strengths show a good agr eement with the true values. (C) 2000 Academic Press.