Well Log Curve Digitization via Neural Network Segmentation

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing well log data is often stored in paper form, making it difficult to digitize and integrate into modern drilling processes.

Innovation Solution

A machine learning method that involves scanning well log paper documents, using a trained segmentation neural network to create a curve mask, and then employing a trained digitization neural network to produce a digitized version of the curve mask.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If well log data is stored in paper form, then data can be preserved and archived, but it becomes difficult to digitize and integrate into modern drilling processes

Engineering Contradiction:
Improvedata preservationVSAvoiddata integration
Core Design Contradiction:
ReliabilityVSEase of operation

Solution Approach 1:

The patent uses optical scanning to create digital copies of paper well log documents. The segmentation neural network processes these scanned images to extract curve information, effectively copying data from physical paper form into digital format that can be integrated into modern drilling processes while preserving the original paper archives

Inventive Principle:
Principle #26Copying

Solution Approach 2:

The patent replaces manual digitization methods (mechanical processes of manually transcribing data from paper to digital format) with an automated neural network system. The segmentation network and coordinate extraction network automatically convert scanned paper images into structured digital data, eliminating the need for manual mechanical digitization work

Inventive Principle:
Principle #28Mechanics substitution (Replace mechanical system)

2Ease of manufacture

If manual digitization methods are used, then data can be converted to digital format, but the process is time-consuming and labor-intensive

Engineering Contradiction:
Improvedigitization processVSAvoiddigitization speed
Core Design Contradiction:
Ease of manufactureVSProductivity

Solution Approach 1:

The patent implements a self-service automated system where the neural networks independently perform the entire digitization process. The segmentation network automatically identifies and segments curves from scanned documents, and the coordinate extraction network automatically converts these segments into digital coordinate data without requiring manual intervention at each step, thereby dramatically improving productivity

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent introduces scanned image documents as an intermediary format between paper well logs and final digital data. The neural networks process these intermediary scan images to extract curve information, creating a bridge that enables automated conversion from physical paper to structured digital format while maintaining data accuracy

Inventive Principle:
Principle #24Intermediary (Mediator)

3Loss of information

If traditional digitization approaches are used, then data can be captured, but accuracy and precision in curve representation may be compromised

Engineering Contradiction:
Improvedata accuracyVSAvoidcurve coordinate precision
Core Design Contradiction:
Loss of informationVSMeasurement precision

Solution Approach 1:

The patent applies segmentation by dividing the well log document into distinct regions using the segmentation neural network. This network identifies and separates individual curves, grid lines, and other elements, ensuring that curve data is extracted with high precision without contamination from surrounding elements, thereby maintaining both completeness and accuracy of the digitized data

Inventive Principle:
Principle #1Segmentation

Data Source

PatentEP4542508A1Well log curve digitization
Publication Date: 2025.04.23 SERVICES PETROLIERS SCHLUMBERGER SA
  • EP4542508A1 patent drawingFigure 1
  • EP4542508A1 patent drawingFigure 2
  • EP4542508A1 patent drawingFigure 3

AI summary

Techniques for digitizing a well log are presented. The techniques include: obtaining a scan of a well log curve paper document; passing the scan of the well log curve paper document to a trained segmentation neural network, such that a curve mask is obtained; passing the curve mask to a trained digitization neural network, such that a digitization of the curve mask is obtained; and outputting the digitization of the curve mask.