Deep Learning Casing Anomaly Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional automated data interpretation techniques for electromagnetic logging in oil and gas wells often lead to false metal loss calculations due to failure in distinguishing actual metal loss from changes in casing pipe properties or downhole completion effects, resulting in uncertain measurement results and low resolution.
Innovation Solution
A computer-implemented method using a deep learning model, specifically a U-Net classifier with convolutional neural networks, is applied to classify anomalies in electromagnetic survey data, differentiating between actual metal loss and other factors by analyzing shape, intensity variation, and vertical extent of anomalies, thereby improving the accuracy of metal loss detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional automated data interpretation techniques are used for electromagnetic logging, then processing speed is improved, but measurement precision deteriorates due to false metal loss calculations
Solution Approach 1:
The patent introduces an intermediary classification system that acts as a mediator between automated processing and final interpretation. A trained machine learning model classifies anomalies into different categories (e.g., actual metal loss vs. other factors like casing deformations or tool effects), allowing automated processing to proceed efficiently while maintaining precision by filtering out false positives through the classification layer
Solution Approach 2:
The patent changes the parameter of anomaly characterization by using multiple features beyond simple signal amplitude, including shape, intensity variation, vertical extent, and morphological patterns. This multi-parameter approach enables the system to distinguish between true metal loss and other anomalies, resolving the contradiction between speed and precision
2Productivity
If conventional automated interpretation is used, then processing efficiency is improved, but reliability deteriorates due to uncertain measurement results
Solution Approach 1:
The patent implements feedback through the machine learning classification system that provides confidence levels and categorical feedback on each detected anomaly. The system feedbacks whether an anomaly is likely true metal loss or false positive, enabling operators to prioritize interpretations with high reliability while maintaining efficient processing of all data
Solution Approach 2:
The classification model serves as an intermediary reliability filter that processes all anomalies automatically but flags only those with high confidence as true metal loss. This intermediary layer maintains processing efficiency while improving reliability by providing certainty levels for each interpretation
3Measurement precision
If detailed analysis of each anomaly is performed manually, then measurement precision is improved, but loss of time increases
Solution Approach 1:
The patent applies preliminary action by pre-training the machine learning model on labeled data containing various anomaly types. This preliminary training enables the system to automatically perform detailed analysis equivalent to manual expert review, achieving high classification accuracy without requiring time-consuming manual examination of each new anomaly
Solution Approach 2:
The patent uses copying by training the machine learning model on copies of expert-labeled data. The model learns from replicated examples of true metal loss and false positive anomalies, enabling it to automatically replicate expert-level classification accuracy without requiring actual manual review time for each case
Applied Scientific Principles
This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.
Function Achieved in This Case
The deep learning model enables accurate and automatic classification of inspection logs, reducing false positives and providing a more precise interpretation of metal loss, thus optimizing well integrity evaluation by distinguishing true metal loss from other environmental factors.
Implementation Method 1
An EM logging tool can leverage multiple operating principles to detect changes in casing pipe thickness
Implementation Method 2
The detected change can be translated into a metal loss percentage using signal processing techniques after raw data has been captured
Data Source
AI summary
Implementations provide a computer-implemented method that includes: accessing a first database holding results of interpreting casing integrity, wherein each result provides a first or a second label for a detected anomaly at a depth location of an inspection log that records electromagnetic (EM) survey data of an underground metal casing; accessing a second database holding inspection logs, each recording EM survey data of a corresponding underground metal casing; training a deep learning model configured to classify an input inspection log into the first or the second label; applying the deep learning model to one or more unclassified inspection logs of the second database, wherein the one or more unclassified inspection logs of the second database comprising anomalies; and subsequently classifying the one or more unclassified inspection logs of the second database into either the first label or the second label.