In this paper, two optimized autofocusing metasurfaces (AFMs) with different desired focal distances are designed by using particle swarm optimization (PSO) algorithm. Based on the finite element simulation software COMSOL Multiphysics, the performance of ultrasound transducer (UT) with AFM at different design parameters in Airy distributions (r0,ω) and the bottom thickness (d) of AFM are simulated and analyzed. Based on the simulation data, the artificial neural network model is trained to describe the complex relationship between the design parameters of AFM and the performance parameters of UT. Then, the multiobjective optimization function for AFM is determined according to the desired performance parameters of UT, including focal position, lateral resolution, longitudinal resolution and absolute sound pressure. In order to obtain AFMs with the desired performance, PSO algorithm is adopted to optimize the design parameters of AFM according to the multiobjective optimization function, and two AFMs are optimized and fabricated. The experimental results well agree with the simulation and optimization results, and the optimized AFMs can achieve the desired performance. The fabricated AFM can be easily integrated with UT, which has great potential applications in wave field modulation underwater, acoustic tweezers, biomedical imaging, industrial nondestructive testing and neural regulation.