Velocity Model Training With Drilling-Constrained Seismic Iteration
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing velocity models constructed using logging data and seismic horizon data struggle to accurately capture geological changes near wells in areas with strong heterogeneity of underground media and a small number of adjacent wells, resulting in low prediction accuracy.
Innovation Solution
A model training method involving pre-processing of drilling and logging parameters, constructing a first velocity model, and performing iterative optimization using seismic data to enhance the model's accuracy, incorporating drilling parameters as constraints and leveraging low-frequency correlations.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If logging data and seismic horizon data are combined to construct velocity model, then the model construction is simple and easy to use, but the prediction accuracy is low in areas with strong heterogeneity of underground media and scanty logging data
Solution Approach 1:
The patent combines three types of data (logging data, seismic data, and drilling parameter data) into a unified velocity model construction framework. The drilling parameter data is integrated as a new constraint condition alongside traditional logging and seismic data, creating a composite data model that leverages the advantages of each data source while compensating for their individual limitations in heterogeneous areas
Solution Approach 2:
The patent transforms drilling parameters (such as rate of penetration, weight on bit, rotary speed) into velocity constraint conditions through established empirical relationships. This parameter transformation allows drilling data to be converted into meaningful velocity constraints that can directly guide velocity model construction and optimization in areas with limited logging coverage
2Measurement precision
If more data sources are integrated to improve prediction accuracy, then the velocity model accuracy improves, but the data processing complexity increases
Solution Approach 1:
The patent divides the velocity model construction process into distinct stages: initial model construction using logging and seismic data, followed by iterative optimization using drilling parameter constraints. This segmentation allows each data type to be processed and integrated in a systematic manner, reducing the overall complexity compared to simultaneous multi-parameter optimization
Solution Approach 2:
The patent implements an iterative optimization mechanism where the velocity model is continuously refined using drilling parameter feedback. The drilling parameters serve as real-time constraints that guide model adjustments, creating a closed-loop system that automatically improves accuracy without requiring manual intervention in the complex multi-parameter optimization process
Data Source
AI summary
The present application provides a model training method and an apparatus, relating to the field of model training technologies. The model training method includes: pre-processing a drilling parameter of an explored well, a logging parameter of the explored well, and a drilling parameter of a target drilling well; constructing a first velocity model according to the pre-processed logging parameter of the explored well; obtaining a predictive drilling parameter according to the pre-processed drilling parameter of the target drilling well and the pre-processed drilling parameter of the explored well; and performing a first-level iterative optimization on the first velocity model according to the predictive drilling parameter and seismic data of the explored well and the target drilling well.