Dynamic characteristic of a system have to be obtained so as to offer the reliable data for the system dynamic modification and dynamic optimum design. The system modal parameters identification can be changed into the global optimization problem in wavelet plane because the wavelet ridges can carry much information of system characteristic parameters. Firstly, the paper will introduce the so-called PSO (Particle Swarm Optimization) technique into the modal analysis and propose the PSO wavelet ridge extraction algorithm. Furthermore, the formulas identifying modal parameters from a wavelet ridge are derived and computation steps were given. The simulation experiments are implemented in order to evaluate precision of this method and the robustness of noise. Finally, the proposed method applies to the modal parameters identification of a watermelon to explore the effectiveness. The experiment results prove the method is precision and insensitive to noise.