Filament Abnormality Detection Using Guide Imaging and Autoencoders
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing methods struggle to accurately detect yarn path abnormalities in filaments due to their fine nature, making visual inspection difficult and unreliable.
Innovation Solution
A filament abnormality detection device using a camera and machine learning-based estimation model, specifically an autoencoder, to analyze filament images and detect deviations from a guide, with preprocessing to enhance image clarity and reduce overexposure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If visual inspection is used to detect yarn path abnormalities, then the detection method is simple, but the detection accuracy is low due to the fine nature of filaments
Solution Approach 1:
The patent replaces the mechanical/visual inspection system with an optical imaging system combined with machine learning algorithms. A camera captures images of filaments passing through the guide, and an estimation model processes these images to detect abnormalities, substituting human visual inspection with automated optical-mechanical-system.
Solution Approach 2:
The patent introduces an estimation model as an intermediary between the camera and the final detection result. This machine learning model processes the raw image data and translates it into actionable detection results, serving as a mediator that bridges the gap between simple imaging and accurate abnormality detection.
2Reliability
If machine learning-based detection is implemented, then detection accuracy is improved, but device complexity increases due to additional processing requirements
Solution Approach 1:
The patent performs preliminary actions by pre-training the estimation model using training images captured during normal operation. This preprocessing step prepares the system in advance, allowing it to reliably detect abnormalities without requiring complex real-time analysis during actual detection.
Solution Approach 2:
The estimation model serves itself by learning from normal operation data and automatically adapting to detect abnormalities. The system uses its own operational data for training, eliminating the need for external expertise or complex manual configuration, thereby improving reliability while managing complexity.
3Measurement precision
If image processing is enhanced to improve detection accuracy, then abnormality detection capability is improved, but processing time increases
Solution Approach 1:
The patent applies partial action by focusing image processing only on relevant features within the captured images. The estimation model is trained to identify specific patterns indicating abnormalities without processing every detail of the image, thereby maintaining high detection precision while reducing overall processing time.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
Provided is a technique for detecting a yarn path abnormality regarding a filament with respect to a guide (11). A filament abnormality detection device detects an abnormality with respect to a spinning take-up device (1) configured to take up a plurality of filaments spun from a spinning device. The spinning take-up device (1) includes a guide (11) for guiding the plurality of filaments fed from upstream to downstream and bringing them closer to each other. The filament abnormality detection device includes a control device (101). The control device (101) executes: processing for acquiring, from a camera (30) arranged so that a shooting range thereof at least includes the plurality of filaments passing through the guide (11), a filament image for estimation in which a plurality of filaments appear; processing for determining whether or not a yarn path abnormality has occurred, based on an estimation model (124) subjected to machine learning so as to detect a yarn path abnormality in which at least one of the plurality of filaments is out of the guide (11), and the filament image for estimation; and processing for outputting a determination result of the processing for determining.