ML Tow End Detection in Composite Structures
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for detecting tow ends in composite structures are inefficient, often requiring manual oversight and are susceptible to errors due to variations in ply orientation and misidentification of shadows or features as tow ends, leading to unreliable results.
Innovation Solution
A machine learning model, specifically a convolutional neural network, is trained to classify pixels in images of composite structures, using a training set of images that include both upper and lower plies, and their classifications, to accurately identify tow ends by accounting for expected locations and orientations, reducing false positives from shadows or features.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If conventional methods are used to detect tow ends, then the detection process can be performed, but the reliability and accuracy of detection deteriorates due to false positives from shadows or features and susceptibility to ply orientation variations
Solution Approach 1:
The system performs preliminary actions by receiving expected tow end locations from a database before conducting the actual detection. This allows the machine learning model to compare detected features against predetermined expectations, improving reliability by filtering out false positives from shadows or unrelated features that wouldn't match the expected locations and orientations.
Solution Approach 2:
The machine learning model transforms the detection problem by changing parameters from simple image pixel analysis to classified probability maps indicating likelihood of tow end presence. This parameter transformation enables the system to account for ply orientation variations and distinguish true tow ends from artifacts, thereby improving both reliability and precision simultaneously.
2Measurement precision
If manual oversight is used to verify tow end detection, then detection accuracy can be improved, but productivity and automation level deteriorate due to required human intervention
Solution Approach 1:
The system implements self-service by using the machine learning model to automatically perform both detection and verification functions that would traditionally require manual oversight. The model classifies pixels and generates probability maps autonomously, enabling the system to maintain high detection accuracy while operating fully automatically, thus improving productivity without sacrificing precision.
Solution Approach 2:
The patent replaces the mechanical system of manual visual inspection with an automated machine learning-based detection system. This substitution maintains or improves detection accuracy through consistent algorithmic analysis while dramatically increasing productivity by eliminating the bottleneck of human review, allowing high-volume automated detection.
3Extent of automation
If contrast-enhanced image analysis is used to identify tow ends, then detection can be automated, but reliability deteriorates due to misidentification of shadows or features as tow ends
Solution Approach 1:
The system receives expected tow end locations and orientations from a database as preliminary information before performing automated detection. This allows the machine learning model to validate detected features against predetermined expectations, maintaining full automation while improving reliability by filtering out false positives from shadows or features that don't match expected characteristics.
Solution Approach 2:
The machine learning model incorporates feedback mechanisms by comparing detected features against expected tow end locations and orientations stored in the database. This feedback loop allows the automated system to correct for false positives, as detected features are validated against predetermined expectations, thereby maintaining automation while significantly improving detection reliability.
Data Source
AI summary
A method of inspecting a composite structure formed of plies of tows is provided. The method involves receiving an image of an upper ply overlapping lower plies, the upper ply tow ends defining a boundary between plies, and applying extracted sub-images to a trained machine learning model to detect the upper or lower ply. Probability maps are produced in which pixels of the sub-images are associated with probabilities the pixels belong to an object class for the upper or lower ply. The method may also involve transforming the probability maps into reconstructed sub-images, stitching together a composite image, and applying the composite image to a feature detector to detect locations of tow ends of the upper ply. The method may also involve comparing the locations to as-designed locations of the tow ends, inspecting the composite structure, and indicating a result of the comparison.