Inversion of Rayleigh wave dispersion curves with an artificial neural network

摘要

The inversion method of Rayleigh wave dispersion curves had been discussed in this paper. At first, the authors studied the influence of elastic parameters on Rayleigh dispersion curves, and also studied the inflexions'movement of vR-λR dispersion curves with parameters' change. And then,the inversion method of artificial neural network was put forward for dispersion curves'inversion, and practical application was used to test the validity of the inversion method. According to the studied results, velocities of transverse waves and thicknesses of media have higher influence on dispersion curves, but densities and velocities of compressional waves have little influence on Rayleigh wave dispersion curves. Because the number of inflexions do not corresponding with the number of interfaces when parameters are changed at times. There is not a theoretic gist to delaminate by inflexions in Rayleigh waves' inversion. The artificial network method can interpret the dispersion curves, and it can be used in practice for the future.