Abstract:Abstract The exploration potential of oil and gas is substantial, with considerable untapped resources available for future exploration and development. This paper investigates the application of time-frequency electromagnetic (EM) methods within the domain of computational electromagnetics, enhanced by deep learning techniques, for the profound exploration of oil and gas reservoirs. To identify zones favorable for hydrocarbon presence, it captures electric and magnetic field responses across time and frequency domains. From these responses, it extracts crucial parameters such as resistivity and excitation polarizability. The methodology initiates with a forward simulation to establish the most productive construction parameters for the time-frequency EM approach. Subsequently, data collected are processed and analyzed to qualitatively evaluate each physical parameter. This is followed by quantitative constraints on resistivity and polarizability, employing the adaptive differential evolution algorithm to execute a comprehensive parameter inversion using the horizontal E-field (Ex) data derived from the time-frequency EM method. This inversion facilitates the determination of resistivity and polarizability distributions at various depths, culminating in the generation of corresponding depth cross-sections. The inversion fitting and example analyses yielded that the apparent resistivity profiles of low-resistance backslope formations and backslope formations with high and low resistance interlayers form relative anomaly magnitudes of about 13% and 7.8%, respectively. In the real measurement from the depression to the uplift zone, the thickness of the giant inclined low-resistance conductive layer gradually decreases. The points of the extreme values and minima gradually move from the low-frequency band to the high-frequency band, which is identical to the simulation law. The anomaly amplitude is between [-0.4 Ω·m, 0.6 Ω·m], which can be accurately imaged. This study innovatively combines deep learning technology and electromagnetism, which provides a basis for the next decision on oil and gas exploration and target evaluation preference.

Modeling extra-deep electromagnetic logs using a deep neural network

Deep Learning-Assisted Real-Time Forward Modeling of Electromagnetic Logging in Complex Formations

An End-to-End Neural Network for Complex Electromagnetic Simulations

Direct multi-modal inversion of geophysical logs using deep learning

The Data Supplement Method of Azimuthal EM LWD Based on Deep Learning

Rapid Surrogate Modeling of Electromagnetic Data in Frequency Domain Using Neural Operator.

Rapid surrogate modeling of magnetotelluric in the frequency domain using physics-driven deep neural networks

EMRNet: End-to-End Electrical Model Restoration Network

Three-Dimensional Magnetotelluric Forward Modeling Through Deep Learning

Deep Learning-based Computational Electromagnetics for Deep Exploration of Oil and Gas Reservoirs

Well Logs Reconstruction Based on Deep Learning Technology

An Intelligent Inversion Method for Azimuth Electromagnetic Logging While Drilling Measurements

Deep learning for predicting permeability logs in offset wells using an artificial neural network at a Waggoner Ranch reservoir, Texas

Deep learning for prediction of complex geology ahead of drilling

Robust machine-learning workflow for subsurface geomechanical characterization and comparison against popular empirical correlations

Physics-Constrained Deep Learning of Geomechanical Logs

Model-Based Synthetic Geoelectric Sampling for Magnetotelluric Inversion With Deep Neural Networks

Real-time forward modeling and inversion of logging-while-drilling electromagnetic measurements in horizontal wells

A new deep-reading look-ahead method in electromagnetic logging-while-drilling using the scattered electric field from magnetic dipole antennas

Generation of Synthetic Density Log Data Using Deep Learning Algorithm at the Golden Field in Alberta, Canada

A Supervised Descent Learning Technique for Solving Directional Electromagnetic Logging-While-Drilling Inverse Problems