Simulation of high-speed train pass-by noise source identification based on MVDR beamforming

To localize train pass-by noise, a hexagonal array that with 37 microphones, has better spatial symmetry, is able to locate noise source in the space of 0-360° and effectively suppress gratinglobes was designed. In order to increase the resolution of hexagonal-array, minimum variance distortionless response (MVDR) is used as beamforming algorithms. The resolution is compared between Conventional beamforming (CBF) and MVDR applied for multiple sources localization. To validate this method, a simple case of multiple sources in close vicinity was investigated by simulation assessment. It is shown that MVDR effectively filters interferences and correctly identifies noise sources compared to CBF. The advantages of hexagonal array and MVDR are discussed and the limitations are also highlighted.