Application of the adaptive array reduction method for offset acoustic source localisation

Abstract

Acoustic beamforming arrays are typically designed assuming that the noise source of interest is aligned with the geometric centre of the array or near it, and with some prior knowledge of the frequency range of interest. This method of array design works well and provides sufficient performance provided that the source of interest is located at or near the centre of the investigated region, known as a scanning grid. If a source is recorded at some distance away from the scanning grid centre, its representation within a beamforming source map can be distorted, the Main Lobe Width (MLW) can be enlarged and the remainder of the source map may possess significant sidelobes. The magnitude of the Maximum Sidelobe Level (MSL) often increases as the source is moved away from the scanning grid centre, which can be problematic when the relative alignment between the array centre and the source cannot be altered. By using an array design that is fixed within a testing facility, the region in which optimal MSL and MLW performance can be obtained is limited. By using a recently published technique known as the Adaptive Array Reduction Method (AARM), these issues can be overcome provided that some information regarding the source location and the frequency range of interest is known prior. An array can be designed for a specific source location and frequency, thus significantly improving the MSL, MLW and lobe-distortion, relative to an array locked in a fixed position. In this paper, the AARM is experimentally verified, revealing excellent agreement with simulated source maps. Furthermore, the AARM is used to design arrays based on a centred and non-centred source and their numerical and experimental performances are compared against some logarithmic spiral and randomised pattern arrays, clearly revealing significantly improved source image maps for acoustic sources located away from the scanning grid centre.