Engineering Component Suitability Assessment Through cGAN Image Discrimination
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods for using artificial intelligence in engineering design and manufacture are limited due to the inability to directly apply engineering analysis to component images, necessitating improved methods for evaluating and assessing manufactured and in-service components.
Innovation Solution
Utilizing a Conditional Generative Adversarial Network (cGAN) with a generator and discriminator, embedding physical parameters as histograms and glyphs within component images, and employing computational fluid dynamic modeling to enhance AI training and assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Extent of automation
If AI systems are used for image recognition in engineering design and manufacture, then functionality and performance can be improved, but the images cannot be directly evaluated using engineering analysis methods
Solution Approach 1:
The patent merges AI image recognition capabilities with traditional engineering analysis methods by integrating multiple assessment modalities (visual inspection, dimensional measurement, material property analysis) into a unified AI system that processes component images comprehensively
Solution Approach 2:
The patent introduces computational models and simulation data as intermediary elements that bridge the gap between AI image analysis and engineering analysis, allowing the system to incorporate stress analysis, thermal analysis, and fluid dynamics predictions into the AI-assessed component evaluations
2Measurement precision
If computational methods are used to evaluate component images for desired performance, then assessment accuracy can be improved, but the process time increases
Solution Approach 1:
The patent performs preliminary actions by pre-training the AI system with extensive computational data and simulation results before actual component assessment, so that during operation the AI can make rapid evaluations without performing time-consuming computational analyses in real-time
Solution Approach 2:
The patent applies partial computational analysis by using AI to perform the most critical and time-sensitive aspects of component evaluation, while reserving detailed computational methods for follow-up analysis of only those components that the AI identifies as potentially problematic
Data Source
AI summary
A computer implemented method of assessing the suitability of a component, including: utilising a Conditional Generative Adversarial Network (cGAN), the cGAN network having an input, a generator and a discriminator; obtaining images or scans of manufactured components for both acceptable and non-acceptable components; training the discriminator of the cGAN with the images representing the acceptable and non-acceptable scans or images of a manufactured component; scanning or imaging a component to be assessed; and inputting the image or scan of a component to be assessed into the discriminator of the cGAN, wherein the output of the discriminator assesses if the scan or image is acceptable for use or not.