In order to resolve the Lamb wave multi-modal aliasing in grounding grid down-conductors due to boundary conditions, dispersion effects, and geometric structures, a hybrid method combining particle swarm optimization (PSO) and orthogonal matching pursuit (OMP) was used to investigate the mechanisms of Lamb wave modal separation and identification in complex environments. The results show that by taking the second-order Taylor expansion of the Lamb wave wavenumber function k(ω) as the core theoretical foundation, the modal separation problem is transformed into a sparse estimation problem for (k₀,k₁,k₂), yielding interpretable parameters consistent with dispersion curves. Meanwhile, PSO is employed to globally search for atoms in the continuous parameter space, thereby avoiding errors associated with traditional discrete dictionaries. Additionally, OMP is utilized to suppress repeated modal selection and energy mismatch under strong aliasing conditions through orthogonalized reconstruction. In simulations and measured signal validations with modal aliasing degrees of 25%, 50%, 75%, and 100%, the method successfully separated aliased modal components such as A₀, A₁, A₂, S₀, S1, and S₂, and accurately extracted their dispersion characteristic parameters, and the estimation accuracy of parameters (k₀,k₁,k₂) is improved by 1-3 orders of magnitude compared to the benchmark methods. It is concluded that this method, through the theoretical correlation between dispersion characteristics and wavenumber expansion coefficients, partially resolves the issue of multi-modal Lamb wave signal analysis. It provides a systematic solution for enhancing the accuracy of grounding grid down-conductor detection and analyzing complex Lamb wave signals.