Unsupervised Machine Learning for Well Integrity Characterization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods lack an effective, deterministic approach to interpret well integrity and cementing properties in cased wells with multiple string casings using high-dimensional sonic data, which is challenging due to the 'curse of dimensionality' and the absence of labeled data for supervised learning.
Innovation Solution
Employing unsupervised machine learning systems, such as Self-Organizing Maps (SOM), Auto-Encoders (AE), Variational Auto-Encoders (VAE), and Generative Adversarial Networks (GAN), to process high-dimensional sonic data and identify low-dimensional representations for clustering, thereby characterizing well integrity without requiring labeled input signals.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If high-dimensional sonic data is used for well integrity characterization, then measurement precision is improved, but device complexity increases due to the curse of dimensionality
Solution Approach 1:
The patent extracts essential features from high-dimensional sonic data by applying dimensionality reduction techniques. The system identifies and extracts key characteristics that define well integrity conditions, separating essential information from redundant dimensions. This extraction process transforms complex high-dimensional data into manageable low-dimensional representations while preserving the critical information needed for accurate well integrity assessment.
Solution Approach 2:
The patent transitions data from high-dimensional space to low-dimensional space through unsupervised machine learning algorithms. By changing the dimensional representation of the sonic data, the system maintains the essential patterns and relationships while reducing computational complexity. This dimensionality change enables effective clustering and interpretation without losing the precision needed for well integrity characterization.
2Ease of operation
If unsupervised machine learning is applied to process sonic data, then ease of operation is improved by eliminating labeled data requirements, but manufacturing precision deteriorates due to lack of supervised guidance
Solution Approach 1:
The patent implements self-service through unsupervised machine learning algorithms that automatically process sonic data without requiring external labeled guidance. The system autonomously identifies patterns, performs dimensionality reduction, and executes clustering based solely on the intrinsic structure of the input data. This self-service approach eliminates the need for manual labeling while maintaining characterization precision through the algorithms' ability to discover meaningful patterns independently.
Solution Approach 2:
The patent replaces manual data labeling and supervised training mechanisms with automated unsupervised learning systems. Instead of requiring expert-labeled datasets to guide the analysis, the system uses automated algorithms to discover patterns and structures in the sonic data. This substitution maintains precision by leveraging the mathematical properties of the data itself rather than relying on external guidance.
3Productivity
If dimensionality reduction is applied to sonic data, then productivity is improved by reducing computational load, but loss of information increases during the transformation from high-dimensional to low-dimensional representation
Solution Approach 1:
The patent extracts essential information from high-dimensional sonic data while discarding redundant dimensions. By identifying and extracting the most significant features that contribute to well integrity characterization, the system reduces computational load without sacrificing critical information. The extraction process focuses on preserving patterns and relationships that are most relevant to the analysis objectives.
Solution Approach 2:
The patent transforms data parameters from high-dimensional to low-dimensional space through unsupervised learning algorithms. This parameter transformation maintains the essential characteristics of the sonic data by preserving the underlying patterns and relationships. The dimensionality reduction process carefully selects which parameters to retain, ensuring that information critical for well integrity assessment is preserved while reducing overall data complexity.
Data Source
AI summary
Methods and systems are provided that characterize and evaluate well integrity of a cased well using unsupervised machine learning of acoustic data. Sonic waveform data for acoustic signals received by the receiver array of a sonic logging tool is collected and processed to determine a high-dimensional representation of the sonic waveform data. The high-dimensional representation is input to an unsupervised machine learning system to determine a low-dimensional representation of the sonic waveform data. A clustering method is applied to the low-dimensional representation to identify a set of clusters therein. At least one well integrity property of the depth interval of the cased well is determined based on the set of clusters. In embodiments, the at least one well integrity property can characterize cement condition in an annulus of the cased well as a function of azimuth and depth and can be used to evaluate cement integrity in a depth interval of the cased well.