Deep Learning Formation Pressure Prediction Using Acoustic Attenuation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Traditional methods for overpressure prediction in oil and gas exploration, based on seismic wave velocity, face challenges due to high gas content and complex structures, leading to multiple solutions and imprecision in pressure prediction.
Innovation Solution
A physical embedded deep learning formation pressure prediction method using the impedance quality factor Q, which replaces the nonlinear activation function of Caianiello convolution neurons with a rock physics model, constructing deep learning convolution neural networks to improve prediction accuracy and efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If traditional seismic wave velocity method is used for overpressure prediction, then the prediction can be performed in overpressure formation caused by disequilibrium compaction, but the prediction accuracy deteriorates due to high gas content causing multiple solutions and strong multi-solution ambiguity
Solution Approach 1:
The patent changes the physical parameter used for pressure prediction from seismic wave velocity to acoustic attenuation (quality factor Q). This parameter substitution resolves the multi-solution ambiguity caused by high gas content, as acoustic attenuation has a stronger and more direct relationship with formation pressure and is less affected by gas saturation variations.
Solution Approach 2:
The patent replaces the traditional mechanical/physical model-based velocity method with a deep learning neural network model that incorporates acoustic attenuation data. This substitution enables the system to handle complex overpressure systems with high gas content by learning non-linear relationships from data, thereby improving prediction accuracy in previously difficult geological conditions.
2Ease of manufacture
If traditional seismic wave velocity method is used, then the method is simple to implement, but the prediction precision deteriorates because seismic velocity data are generally smooth and cannot carry out precision pressure prediction analysis
Solution Approach 1:
The patent introduces acoustic attenuation (quality factor Q) as an intermediary parameter that provides higher resolution and more detailed information about formation pressure variations. This intermediary carries richer geological information compared to smooth velocity data, enabling precision pressure prediction analysis while maintaining computational feasibility through the deep learning framework.
Solution Approach 2:
The patent adds a new dimension of information by using acoustic attenuation characteristics alongside traditional velocity data. The quality factor Q provides complementary information about energy loss and scattering, creating a multi-dimensional feature space that enables more precise pressure prediction through the neural network's ability to process complex multi-parameter relationships.
3Measurement precision
If deep learning Caianiello convolution neural networks (CCNNs) with rock physics model are constructed, then the pressure inversion accuracy and network learning efficiency are greatly improved, but the device complexity increases due to the complex neural network structure
Solution Approach 1:
The patent performs preliminary action by pre-training the deep learning CCNN model using well-logging data to establish accurate rock physics relationships. This pre-training phase prepares the network with prior knowledge of formation characteristics, enabling it to achieve high pressure inversion accuracy when applied to seismic data, thereby reducing the need for complex iterative adjustments during actual prediction operations.
Solution Approach 2:
The patent implements self-service through the deep learning model's ability to automatically learn and extract relevant features from acoustic attenuation data without requiring manual feature engineering. The neural network self-optimizes its internal parameters and structures during training, reducing the need for complex external control systems and simplifying the overall operational complexity despite the sophisticated model architecture.
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
This approach enhances the accuracy of formation pressure prediction by using acoustic attenuation instead of velocity, providing a deterministic rock physics model that improves feature extraction and consistency across neurons, leading to more precise results compared to traditional methods.
Implementation Method 1
characterizes seismic attenuation by impedance quality factor Q
Data Source
AI summary
The present invention discloses a physical embedded deep learning formation pressure prediction method, device, medium and equipment, the present invention characterizes seismic attenuation by logging impedance quality factor Q, based on the Q value and rock physics model of formation pressure, the physical mechanism of this kind of certainty replace Caianiello convolution neurons of the nonlinear activation function, using the convolution neurons, build deep learning convolution neural networks (CCNNs), can greatly increase the stress inversion precision and learning efficiency, get accurate formation pressure prediction results. Compared with the prior art, the present invention uses acoustic attenuation instead of the traditional acoustic velocity to characterize formation pressure, and solves the problem that the traditional pressure prediction method based on velocity has strong multiple solutions due to high gas content and complex structure.