We found a match
Your institution may have access to this item. Find your institution then sign in to continue.
- Title
Seisminės skersinės bangos greičio modeliavimas grunte panaudojus ekstremalųjį apmokymo algoritmą.
- Authors
Urbaitis, Donatas; Indriulionis, Audrius; Mokrik, Robert; Dundulis, Juozas Kastytis
- Abstract
Artificial neural networks are more and more often applied not only in diagnostics, finance, regulation and computer graphics, but also in engineering geology. In our work, artificial neutral networks were used for studying strength characteristics indexes in Lithuanian soils for the first time. Investigation involved analysis of till clayey soil from Stabatiškės area. In the study area, the cone penetration test was carried out for determination of conditional strength of soil, and the seismic cone penetration test was carried out to estimate shear wave velocity (Vs). The extreme learning machine (ELM) was used in artificial neutral training mode, and stochastic search models were used in data optimization, which discovered the values of global maximum and minimum. Differential evolution (DE) algorithm was used in this modelling (simulation). The selected input parameters in the model were: corrected cone resistance (qt), sleeve friction (fs), friction ratio (Rf), vertical geostatic stress (σ), vertical effective stress (σ'), and void ratio (e). Shear wave velocity (Vs) was an output parameter. After modelling, prediction analogues of shear wave velocities were made, which allowed estimating dynamic soil properties without seismic testing.
- Publication
Geologija, 2016, Vol 2, Issue 2, p92
- ISSN
1392-110X
- Publication type
Article