Optimization of Planar Helical Microphone Array Based on Rotation Angle Difference Randomization Strategy of Adjacent Array Elements

Abstract

In this paper, a microphone array for the localization of abnormal noise class defects in dry-type transformers is optimized. Near-field spherical waves are used as the dry transformer acoustic signal model. The array aperture, the number of array elements, the number of spiral arms, the imaging area size, the source frequency, and the distance from the source to the array plane are the constraints. An optimization strategy of randomizing the angular difference of adjacent array elements on the spiral arm is used to construct the solution set. The optimization objectives are the weighted sum of the half-beam width and the maximum sidelobe level is minimized. The solution is performed using the bacterial foraging algorithm. A 64-element array with an aperture of 0.5 m is used as a validation example. The optimized array achieves the best performance at an arm number of 8. The variation pattern of array performance with source frequency and source distance is visually portrayed using a heat map. Compared with the Unerbrink array, which is also designed with an 8-arm spiral structure, the array designed in this paper has a wider detection frequency range and a larger detection range.