A microphone array has been employed to locate the major sound sources on a moving low speed electric train and analyze the characteristics of train noise. Besides using the traditional delay and sum beamforming for analysis, a new deconvolution method was developed which works for a moving line source model similar to the moving train noise. In the noise model, a moving line source was represented by a series of discrete moving sources, with some sources having higher power and the powers of the other sources having small fluctuation about a mean level. This noise model was verified in our in-situ measurement where the major sources were located at the wheel positions and the variation of noise level along the middle part of a train unit were found to be small. Different simulations we...[ Read more ]

A microphone array has been employed to locate the major sound sources on a moving low speed electric train and analyze the characteristics of train noise. Besides using the traditional delay and sum beamforming for analysis, a new deconvolution method was developed which works for a moving line source model similar to the moving train noise. In the noise model, a moving line source was represented by a series of discrete moving sources, with some sources having higher power and the powers of the other sources having small fluctuation about a mean level. This noise model was verified in our in-situ measurement where the major sources were located at the wheel positions and the variation of noise level along the middle part of a train unit were found to be small. Different simulations were performed and results showed that the new deconvolution method can give more accurate source power estimates than the delay and sum beamforming method.

From the train noise analysis, it was found that the wheels are the major sources at low and mid frequency range, but not at high frequency (>l.5kHz). Rail vibration was deduced to be the major source at high frequency range. The relative noise contribution from different parts of a moving train was investigated. Large variation of noise level emitted from wheel positions were found and should be related to their maintenance conditions.