Skip to content

Latest commit

 

History

History
52 lines (39 loc) · 2.65 KB

README.md

File metadata and controls

52 lines (39 loc) · 2.65 KB

Sonic Log Prediction

Field Application of Intelligent Algorithms for Sonic Log Prediction in the Mishrif Oil Formation, Iraq

This repository contains the code, data, and documentation for predicting sonic logs using intelligent algorithms, specifically applied to the Mishrif Oil Formation in Iraq. The project demonstrates the effectiveness of machine learning models, such as XGBoost, Artificial Neural Networks (ANN), and Random Forest (RF), in predicting sonic logs from well log data.

Table of Contents

  1. Introduction
  2. Methodology
  3. Results
  4. Repository Structure

Introduction

Sonic logs, which measure the travel time of sound waves through subsurface formations, are critical for reservoir characterization, geomechanical analysis, and drilling optimization. However, acquiring sonic logs can be expensive and time-consuming. This project leverages machine learning to predict sonic logs from other well log data, such as gamma ray, resistivity, density, and neutron porosity.

The study focuses on the Mishrif Formation, a major carbonate reservoir in Iraq, known for its heterogeneity and complex lithology.

Methodology

Data Preparation

  • Input Data: Well log data including Gamma Ray (GR), Deep Resistivity (LLD), Shallow Resistivity (LLS), Bulk Density (RHOB), and Neutron Porosity (NPHI).
  • Target Variable: Sonic log (DT - compressional wave travel time).
  • Preprocessing:
    • Handling missing values.
    • Normalizing/standardizing data.
    • Splitting data into training and testing sets.

Machine Learning Models

  • Algorithms:
    • XGBoost: A gradient-boosting algorithm known for its accuracy and efficiency.
    • Artificial Neural Networks (ANN): Captures complex, non-linear relationships.
    • Random Forest (RF): An ensemble method suitable for high-dimensional data.
  • Evaluation Metrics:
    • R² (Coefficient of Determination): Measures how well the model explains the variance in the target variable.
    • RMSE (Root Mean Squared Error): Quantifies the prediction error.
    • MAE (Mean Absolute Error): Provides a straightforward measure of error magnitude.

Results

  • XGBoost achieved the highest R² (e.g., 0.85) and lowest RMSE, indicating superior predictive accuracy.
  • ANN performed well but required more computational resources.
  • RF provided robust predictions but was slightly less accurate than XGBoost.
  • Feature Importance: Bulk Density (RHOB) and Neutron Porosity (NPHI) were the most influential features for predicting sonic logs.

Repository Structure