Physics-Guided CNN Workflow for Automated Seismic Inversion
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing seismic inversion workflows are complex, require significant computational resources, and heavily rely on manual supervision, making them inefficient and labor-intensive.
Innovation Solution
A deep learning workflow that utilizes two physics-guided convolutional neural networks (CNNs) to automate the seismic inversion process, integrating seismic data with well logs to estimate geophysical properties with minimal human intervention.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional seismic inversion workflows are used, then accurate geophysical property models can be obtained, but the process requires significant computational resources and manual supervision
Solution Approach 1:
The patent replaces traditional mechanical/computational seismic inversion methods with a neural network-based system. The neural network is trained on synthetic seismic data and well log data, then applies learned patterns to invert actual seismic data into geophysical property models, eliminating complex iterative computational workflows and manual supervision while maintaining accuracy.
Solution Approach 2:
The patent performs preliminary training of the neural network using synthetic seismic data generated from well log data before actual inversion. This pre-training phase creates a ready-to-use model that can directly process field seismic data without requiring complex real-time computational resources or manual intervention during the actual inversion process.
2Productivity
If traditional seismic inversion workflows are used, then geophysical property models can be generated, but the process is labor-intensive and inefficient
Solution Approach 1:
The neural network performs the inversion process autonomously without requiring manual supervision. The system self-services by automatically processing seismic data, applying the trained model, and generating geophysical property models independently, eliminating labor-intensive manual operations and significantly improving efficiency.
Solution Approach 2:
The patent replaces manual labor and traditional computational workflows with an automated neural network system that processes seismic data efficiently, reducing both time requirements and human intervention while maintaining or improving inversion quality.
3Extent of automation
If deep learning workflow with two CNNs is used, then human intervention is minimized and processing is accelerated, but the system requires training data extraction and synthesis
Solution Approach 1:
The patent performs preliminary generation of synthetic seismic data from well log data before training the neural networks. This pre-computed training data enables the CNNs to be trained offline, allowing the actual inversion process to be fully automated without requiring complex real-time data synthesis or manual intervention during deployment.
Solution Approach 2:
The patent introduces synthetic seismic data as an intermediary between well log data and the neural network training process. This intermediary dataset bridges the gap between available well data and the seismic inversion task, enabling automated training and deployment of the inversion system without requiring complex direct mappings or manual data preparation.
Data Source
AI summary
A method for converting seismic data into geophysical property models. The method includes receiving seismic data from a site, wherein the seismic data includes seismic stacks and well logs received from the site. A large-scale structural model (LSSM) is then constructed from the received seismic data. The method includes estimating initial geophysical properties related to the site using a first neural network trained by the seismic data and the constructed LSSM. Data may then be extracted from the received data and used to train a second neural network. The method also includes revising the initial geophysical properties using a second trained neural network. The revised geophysical properties, which may include p-velocity, density, or s-velocity, may be displayed on a screen. The method also includes performing a site action based on the revised geophysical properties.