Neural network model field of the time of the first entrances of seismic waves of Ukraine

摘要

It is proposed to use the seismic data of the DSS on the territory of Ukraine to build neural network models of the seismic signal arrival times field, which allows the construction of longitudinal wave hodographs along arbitrarily oriented profiles in regions where do not have seismic observations. Also this possible use to correct the velocity section using one from methods of the inversion of hodograph. A comparison is made of simulated arrival times of longitudinal waves with those that were recorded on the territory of Ukraine. Parallels have been made between the modeling of the time field from meridional and latitudinal sources with the propagation of stresses, which external indenters create on the territory of Ukraine. At the same time, regional fault zones were reflected in the residuals of the time field. This shows that the artificial field of time, modeled by the neural network, is reflected in the real tectonic division of the territory of Ukraine.