B-Scan Hyperbola Analysis for Automated Subsurface Depth Estimation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing non-destructive testing (NDT) techniques for determining the depth of subsurface objects are error-prone and labor-intensive, relying on manual approximations of subsurface medium properties, leading to inaccurate depth estimates when the medium is heterogeneous.
Innovation Solution
A fully automated process using machine learning and geometric analysis to identify hyperbolic curves in B-Scan images from NDT scanners, extracting geometric features to determine the dielectric constant and subsequently the depth of buried objects, adaptable for real-time or remote execution.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual approximation methods are used to determine subsurface medium properties, then the process is simple to operate, but the measurement precision of depth estimates deteriorates in heterogeneous media
Solution Approach 1:
The patent replaces manual mechanical approximation methods with an automated machine learning system that processes B-Scan images. The system uses trained models to automatically identify hyperbolic patterns, extract geometric features, and calculate depth estimates, eliminating the need for manual intervention while significantly improving measurement precision in heterogeneous subsurface environments.
Solution Approach 2:
The patent introduces an intermediary machine learning processing layer between the raw NDT data and the final depth estimation. This intermediary system automatically determines subsurface medium properties by analyzing hyperbolic curve characteristics in B-Scan images, serving as a bridge that translates complex radar data into accurate depth measurements without requiring manual approximation.
2Productivity
If automated machine learning processing is implemented, then the productivity and inspection rate improve, but the device complexity increases
Solution Approach 1:
The patent applies preliminary action by pre-training machine learning models using synthetic B-Scan images generated from known subsurface configurations. This offline training phase prepares the system in advance, allowing it to perform rapid automated depth estimation during actual inspections without requiring complex real-time adjustments, thereby improving productivity while managing system complexity.
Solution Approach 2:
The patent uses synthetic copies of B-Scan images generated from simulated subsurface environments to train the machine learning models. These synthetic data copies allow the system to learn from a wide variety of scenarios without requiring extensive field data collection, enabling high productivity in real inspections while keeping the actual field equipment relatively simple.
Data Source
AI summary
A method for estimating a depth of one or more buried objects can include receiving, from a non-destructive testing (NDT) scanner, a B-Scan image of a scanned region. Using a machine learning model, a hyperbola in the B-Scan image can be identified. At least one geometric feature of the hyperbola can be determined. Based upon the at least one geometric feature of the hyperbola, a dielectric constant in at least a portion of the scanned region can be determined. A depth of the one or more objects beneath a surface of the scanned region can be determined based upon the determined dielectric constant.