engleski

Employing machine learning algorithm for cross validating porosity-velocity model

As the technology and acquisition methods of geophysical exploration have developed, the data acquired have also become more complex in its volume. While obtaining data from various sources has become more accessible in some ways, the need to enhance interpretation methods has become even more important. In this context, machine learning has emerged as a solution for highlighting hidden or unknown relationships between different scale data. The relationship between 3D seismic volume and well data could be effectively analyzed using the Self-organizing maps (SOM) algorithm in an unsupervised learning process. The self-organizing map is an artificial neural network that emphasizes patterns in samples by clustering and classifying them into various sets. It has been proven successful in accentuating information about geological features and predicting missing attributes in different geophysical parameters. The study area is located in the Drava Depression, which is known for extensive surveys conducted. The SOM algorithm was used to create low- dimensional maps in order to reduce the dimensionality of 3D seismic data and to predict missing acoustic and density logs based on data measured in a few boreholes. This method has been utilized to produce maps of porosity distribution and velocity models in the study area.

Self-organizing maps ; 3D seismic data ; well logging data ; Drava Depression

nije evidentirano