Application of Genetic Algorithm in Inversion of Raighly Waves Dispersion Curves of Array Measurement of Microtremores (Technical Note)

Inversion analysis of dispersion curves for surface waves is one of recent and important applications in surface soil explorations. This is a challenging problem for linear inversion procedures to be used in evaluation of surface soil mechanical properties due to a highly non-linear nature of soil and to the possibility of large numbers of local minima and maxima of the objective function (multi-modality) in any inversion procedure. As the important outcome of the current study, Genetic algorithm (GA) method is utilized in optimizing the inversion of measured dispersion curves. The main objective of application of Genetic Algorithm in this study is to estimate unknown shear-wave velocity profile of soils in such a manner that the most conformity achieves between experimental dispersion curve obtained by Array measurements of microtremors and the theoretical dispersion curve obtained by the relations which are presented by Haskell (1953) based on wave-propagation theory in layered medium. GA firstly assumes random values for unknown parameters and then by using Selection, Crossover, Mutation and Elitism operators and producing new estimation for these parameters improves the results so that the results in each generation becomes more accurate. The effectiveness of the presented method is considered by an example presented in this study. The results of provided example show acceptable coincidence between the theoretical and experimental dispersion curves. Also, the shear-wave velocity profile of the example site obtained by the new method agrees well with the profiles which are obtained by another alternative method of determination of shear-wave velocity profile.