In order to break through the challenge of high-performance detection for small and medium-sized autonomous underwater vehicles (AUVs) under the constraints of platform size and load capacity, this paper proposes the design idea of foldable sonar adapted to small and medium-sized AUVs to solve the problem of contradiction between small platforms and large-aperture arrays. The main focus of this paper is to propose a sonar formation optimization method based on a multi-objective particle swarm algorithm to achieve high-performance detection while reducing the peak side lobe level (PSLL) and obtaining the narrowest possible mainlobe beamwidth (MBW). Based on the analysis of simulation experiments, an experimental prototype of a foldable sonar array suitable for a certain type of small and medium-sized AUVs was built, and tank experiments were carried out to verify that the optimized array formation has better azimuth estimation performance.