Turbomachine Component Inspection Using X-Ray Metadata Classification
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for inspecting turbomachine components for defects, such as cracks and corrosion, are not sufficiently reliable and require manual intervention despite the availability of software solutions.
Innovation Solution
A method utilizing a machine learning system, trained with images and metadata, to automatically classify turbomachine components as 'servicable' or 'non-servicable', and optionally 'repairable' or 'non-repairable', based on X-ray or CT scans, with the ability to learn from human feedback and adjust evaluation criteria.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated software solutions are used for defect detection, then productivity is improved, but reliability deteriorates due to insufficient accuracy requiring manual inspection
Solution Approach 1:
The patent replaces manual mechanical inspection with an automated machine learning-based image analysis system. The system uses trained algorithms to automatically detect and classify defects in turbomachine components from images, eliminating the need for manual visual inspection while maintaining high reliability through sophisticated pattern recognition capabilities.
Solution Approach 2:
The machine learning system performs self-learning and self-improvement by continuously processing inspection images and refining its defect detection algorithms. The system automatically classifies defects without human intervention, making autonomous decisions about component serviceability based on learned patterns from training data.
2Reliability
If manual inspection is performed to ensure reliability, then defect detection accuracy is improved, but productivity deteriorates due to time-consuming processes
Solution Approach 1:
The patent replaces manual inspection processes with automated machine learning-based image analysis. The system processes images rapidly using computational algorithms, achieving both high speed and high accuracy simultaneously by eliminating the inherent trade-off between manual thoroughness and automated speed.
Solution Approach 2:
The machine learning system acts as an intermediary between image acquisition and final inspection decisions. It processes images through trained models that have learned to recognize defect patterns, providing automated classification results that maintain reliability while dramatically improving inspection throughput.
3Measurement precision
If comprehensive metadata is collected for component classification, then measurement precision is improved, but device complexity increases due to additional data acquisition requirements
Solution Approach 1:
The machine learning system is designed to handle multiple types of input data (images, metadata, operational parameters) through a unified processing framework. This multi-functional approach allows the system to accept diverse data sources without requiring separate specialized processing paths, managing complexity while maintaining comprehensive analysis capabilities.
Solution Approach 2:
The patent segments the data processing into distinct modules: image acquisition, metadata collection, feature extraction, and classification. This segmentation allows each component to be optimized independently while working together in an integrated system, reducing overall complexity through modular design.
Data Source
Figure 1~2
Figure 3~4
AI summary
The present invention relates to a method for checking a component, in particular a component of a turbomachine (1), comprising the steps of: - taking (S2) at least one X-ray image or CT image of the component (10) using an image capturing device (20); - providing (S21) metadata to the component (10), which include in particular the type of component, an operational life of the component (10), a number of remaining life cycles and/or a repair history; - classifying the component (10) based on the image taken by the image capturing device (20) and the provided metadata into a "Servicable" or a "Non-Servicable" category by means of a machine learning system (30).