<p>The inversion of Rayleigh wave dispersion curves faces challenges such as multi-parameter, nonlinearity, and multi-solution. In addition, traditional dispersion curve inversion methods can only obtain two sets of parameters that are shear wave velocities and layer thicknesses. However, using preset shear wave velocities, compressional wave velocities, densities, and layer thicknesses as prerequisites for inversion can easily introduce systematic errors. This may lead to insufficient inversion accuracy that will inevitable seriously restrict the ability to use Rayleigh waves to carry information and solve geological problems. To this end, a method called Adaptive and Spiral Exploration Artificial Fish Swarm Algorithm (ASE-AFSA) was proposed for the inversion of Rayleigh wave dispersion curves. In this algorithm, we used four sets of parameters including shear wave velocities, compressional wave velocities, densities, and layer thicknesses as the target parameters for inverting Rayleigh wave dispersion curves. We also introduced L1 norm regularization into the objective function to constrain the density parameters in order to effectively improve the inversion accuracy and enhance the sparsity of the solution. The ability to balance global exploration and local optimization was achieved through an adaptive field of view and step size adjustment mechanisms. And the ability to traverse in high-dimensional parameter spaces has effectively enhanced by introducing a spiral exploration strategy. In addition, the crowding factor constraints was ignored to improve convergence speed and inversion accuracy. The experimental results of two typical geological models had verified the correctness of the method. Our method could still maintain the stability of inversion in the presence of 10% random noise interference, and its convergence speed and inversion accuracy were superior to other globally optimized nonlinear inversion algorithms such as firefly algorithm. Our method was further applied to an actual seismic dataset processing, and the geological structure characteristics of the survey area were obtained. The research results indicate that our proposed method can not only generate high-precision shear wave velocity models, but also effectively derive compressional wave velocity and density models, thereby offering a robust inversion technique for complex near-surface geological surveys.</p>