APPLYING MPSO FOR BUILDING SHEAR WAVE VELOCITY MODELS FROM MICROTREMOR RAYLEIGH-WAVE DISPERSION CURVES

作者：Zarean A^*; Mirzaei N; Mirzaei M

来源：Journal of Seismic Exploration, 2015, 24(1): 51-82.

• particle swarm optimization
• mutation
• Rayleigh-wave
• dispersion curves
• microtremor

Surface waves have been increasingly used as an attractive tool for obtaining near-surface shear-wave velocity profiles. Inversion of surface wave dispersion curves is a challenging problem for most of local-search methods due to their high nonlinearity and multimodality (large numbers of local minima and maxima of the misfit function). Rayleigh- and Love-wave dispersion curves derived from microtremor arrays have the advantage of not requiring artificial sources; however, they have disadvantages of high uncertainty, low sampling number and limited frequency band. Among many approaches which have been proposed for surface wave inversion thus far, metaheuristic algorithms have been effectively applied to solve it, and avoid trapping in local minima. In this study, a hybrid approach was proposed for inversion of surface wave dispersion curves. The method is a genetic algorithm mutation based particle swarm optimization, namely MPSO. The mention for using the additional mutation operator in this study was to prevent early convergence on local optima of the solution. In each iteration, a hybrid mutation scheme was applied to search the neighborhood area of the solution which corresponds to the two best particles: the best current particle and the best particle found so far. The population was divided into two parts; the first one was regenerated according to the particle swarm optimization and the latter was generated by applying the proposed here. In this work, in order to invert the dispersion curves, a new MATLAB code was developed for the MPSO algorithm. Also, to evaluate calculation efficiency and MPSO stability for inversion of surface wave data, various synthetic dispersion curves were inverted. Following this stage, a comparative analysis with the original particle swarm optimization and the genetic algorithm was made. Consequently, using the MPSO algorithm, the Rayleigh-wave dispersion curve was inverted and one-dimensional V-s profiles were obtained. In conclusion, the proposed approach represents an improvement of a purely particle swarm optimization scheme and the MPSO typically offers a more significant and precise solution in the case of synthetic models. Results of inversion performed on a field data set were validated by borehole stratigraphy.

• 出版日期2015-2