Deep Learning Model for Electrical Imaging Well Logging Facies Recognition
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for automatically recognizing electrical imaging well logging facies are subjective, have low recognition accuracy, and are limited in application range, failing to meet the needs of oilfield production and geological research.
Innovation Solution
A deep learning method is employed to acquire, preprocess, and analyze historical well logging data, constructing a deep learning model with multiple hidden layers to recognize and predict well logging facies, achieving high accuracy and efficiency in reservoir distribution prediction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If traditional expert-based interpretation is used, then subjective judgment is avoided, but interpretation efficiency is low and cannot meet urgent oilfield production needs
Solution Approach 1:
The patent replaces the mechanical expert interpretation system with an automated deep learning system. The deep learning model automatically extracts features from electrical imaging well logging images and classifies facies types, substituting human experts' manual analysis with an automated computational system that processes data much faster and without subjective bias.
Solution Approach 2:
The deep learning model is trained on historical well logging data and then autonomously performs facies recognition without requiring continuous human intervention. Once trained, the system serves itself by automatically processing new imaging data, extracting features, and generating facies classification results independently.
2Measurement precision
If image segmentation and fuzzy mathematics methods are used, then automatic recognition is achieved, but recognition accuracy is low due to control by segmentation quality
Solution Approach 1:
The patent extracts key geological features directly from the electrical imaging well logging images using deep learning convolutional layers, bypassing the need for traditional image segmentation. The deep learning model automatically identifies and extracts relevant features such as sedimentary structures, textures, and patterns without requiring manual or algorithmic segmentation, thereby avoiding the accuracy limitations imposed by segmentation quality.
Solution Approach 2:
The patent transforms the approach by changing from segmentation-based parameters to deep learning feature parameters. Instead of relying on segmentation quality metrics, the system uses learned feature representations from multiple convolutional layers that capture essential geological patterns, fundamentally changing the parameter space and achieving higher recognition accuracy.
3Adaptability or versatility
If template matching methods are used, then recognition is simplified, but application range is narrow and recognition accuracy is low
Solution Approach 1:
The deep learning model is designed to be universal and adaptable to various well logging facies types and geological conditions. By training on diverse historical data encompassing multiple facies categories and geological environments, the model learns generalizable features that enable it to accurately recognize different facies types across various applications, unlike template matching which is limited to pre-defined templates.
Solution Approach 2:
The patent performs preliminary training of the deep learning model on extensive historical well logging data before deployment. This preliminary action of training the model on diverse data sets enables it to adapt to various facies types and geological conditions, expanding its application range and improving its recognition accuracy across different scenarios before it is actually used for recognition tasks.
Data Source
AI summary
A method and an apparatus for automatically recognizing an electrical imaging well logging facies, wherein the method comprises: acquiring historical data of electrical imaging well logging; pre-processing the historical data of the electrical imaging well logging to generate an electrical imaging well logging image covering a full hole; recognizing a typical imaging well logging facies in the electrical imaging well logging image covering the full hole, and determining the electrical imaging well logging image covering the full hole as a training sample in accordance with a category of the imaging well logging facies; constructing a deep learning model including an input layer, a plurality of hidden layers, and an output layer; training the deep learning model using the training sample to obtain a trained deep learning model; and recognizing the well logging facies of the electrical imaging well logging image of the well section to be recognized using the trained deep learning model.