Применение нейронных сетей при интерпретации данных электромагнитных зондирований вертикально-трещиноватых сред
Московский Игорь Георгиевич, Oleg Balaban, Федорова Ольга Сергеевна et al. · Интернет-журнал Науковедение · 2015
In article possibility of application of artificial neural networks for definition of existence in a geoelectric section of a non-uniform layer with frequency dispersion of the conductivity caused by Maxwell-Wagner's effect according to magnetotelluric soundings is investigated. The neural network qualifier of the magnetotelluric data calculated for the three-layer environment with a non-uniform vertically jointed second layer is considered. Two classes of geological environments are allocated, to one of which the qualifier has to correlate magnetotelluric data: Wednesday, containing vertically jointed layer and Wednesday with uniform layers. Two-layer and three-layer neural networks of direct distribution are taken as a basis of the neural network qualifier. In article the description of algorithm of construction and training of the neural network qualifier is given, results of an assessment of quality of recognition the iskusvennykh of neural networks of the simulated data are given.
Cited by 0 works.