Implementations of microphone arrays for railway noise identification

Abstract

The noise from railway operations mainly consists of traction noise, aerodynamic noise and rolling noise, and it is often in the interest of track and vehicle engineers to keep the noise levels below certain limits. To achieve effective noise control for railway, it is essential to evaluate the contributions from all the main components. Therefore, it is necessary to have means and procedures that have the capability to distinguish the contribution of each noise components based on practical measurements. Microphone arrays combined with beamforming algorithms provide a possible way to achieve source identification. Various approaches based on microphone arrays and beamforming for railway noise identification are proposed and extended in this thesis.

The beamforming procedure for static sources based on delay-and-sum is introduced first. Two parameters, array resolution and maximum side lobe level, which characterise the performance of an array configuration, are extracted from the beamforming results in the form of a visualised acoustic map. These two parameters are utilized to determine the objective functions of the Genetic algorithm that is adopted for the optimization of array configurations.

Subsequently, the classic beamforming algorithm for moving sources is presented together with a new method specifically developed having the railway application as a first target. This new method is developed based on a linearisation of the trajectory of the moving sources with a short time window. Both the beamforming algorithms are verified through numerical simulations for moving source. It is demonstrated that the new method has the benefical feature of reducing the side lobe levels of the beamforming results.

Moreover, the influence of source directivity on the beamforming maps and the quantification results is studied and it is shown that the basic assumptions of incoherent monopoles adopted for beamforming is usually not satisfied for railway noise sources. Additionally, the rail noise radiation, which has particular directional properties is studied separately. The relationship between the rail structural waves and their beamforming responses is established based on numerical modelling and simulations. It is found that beamforming algorithms can only capture the noise radiated from the vicinity of the excitation and suppressing the contribution from the propagating structural waves. This leads a misleading results for wheel contributions.

Finally, the results obtained from three measurement campaigns are presented by adopting the methods developed and extended in this thesis. The TWINS models are also introduced to assist the analysis of the results. The procedure presented are used to evaluate a method to separate railway noise sources. Although feasible for the wheel component, it has limitations for the rail. Generally, these results show that the approaches given in this thesis have the capability of giving reasonable estimation for the railway noise sources.

More information

Identifiers

ORCID for Giacomo Squicciarini: orcid.org/0000-0003-2437-6398

Catalogue record

Export record