Developing of New Facets of Indirect Modeling in the Geosciences

TL;DR

This paper explores advanced indirect modeling techniques in geosciences, applying neuro-fuzzy systems, rough sets, and complex networks to analyze permeability, lost circulation, and rock joint behavior, revealing key attributes and rules.

Contribution

It introduces the combined use of SONFIS, SORST, and complex network methods for improved geoscientific modeling and analysis.

Findings

Effective rule extraction for permeability analysis

Identification of dominant system attributes

Complex network modeling of rock joint behavior

Abstract

In this paper, we describe some applications of Self Organizing feature map Neuro-Fuzzy Inference System (SONFIS) and Self Organizing feature map Rough Set (SORST) in analysis of permeability at a dam site and lost circulation in the drilling of three wells in Iran. Elicitation of the best rules on the information tables, exploration of the dominant structures on the behaviour of systems while they fall in to the balance of the second granulation level (rules) and highlighting of most effective attributes (parameters) on the selected systems, are some of the benefits of the proposed methods. In the other process, using complex networks (graphs) theory - as another method in not 1:1 modelling branch- mechanical behaviour of a rock joint has been investigated. Keywords: Information Granules; SONFIS; SORST; Complex Networks; Permeability; Lost Circulation; Mechanical Behavior of a Rock Joint