Anzeige
Erschienen in:
2021 | OriginalPaper | Buchkapitel
Estimating Spatial Correlation Statistics from CPT Field Data, Using Convolutional Neural Networks and Random Fields
verfasst von : Jonathan D. Nuttall
Erschienen in: Challenges and Innovations in Geomechanics
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
In site characterization and modelling, subsurface spatial variability is often characterized using the scale of fluctuation, Θ, in the horizontal and vertical directions. Typically, these scales are estimated by statistically fitting an appropriate autocorrelation function to the CPT data from the layer of interest. These statistical techniques are data intensive requiring significant amounts of data to provide accurate estimates. While in the vertical direction, along the CPT, the available data is abundant, the amounts of data in the horizontal plane is limited to the number of CPTs undertaken; therefore, these traditional approaches can adequately estimate vertical scales, Θv, while the horizontal estimate, Θh, is difficult to obtain.
This paper aims to expand on the previous work of the Author, in using a neural network-based approach to estimate these spatial statistics from CPT data. This study expands the work from simple 2D domains, modelled as random fields, to 3D, and more realistic site surveys, and the methods are compared.