False Twist Component Wear Detection Using Multi-Scale Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The frequent wear of false twist components in texturing machines affects the texturing process flow, necessitating an automated monitoring solution.
Innovation Solution
A fault detection method involving image capture, attention modules, encoder and decoder networks, and multi-layer perceptrons to automatically detect the condition of false twist components in texturing machines, utilizing fine- and coarse-granularity features and historical data for accurate fault detection.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If automated monitoring of false twist component is implemented, then productivity is improved, but device complexity increases
Solution Approach 1:
The patent replaces manual inspection methods with an automated image-based detection system. An image acquisition device captures images of the false twist component, and a detection model processes these images to identify wear and faults, substituting mechanical/manual monitoring with an automated vision system that improves productivity while managing complexity through software-based solutions.
2Measurement precision
If fault detection accuracy is improved through multiple processing stages, then measurement precision is improved, but device complexity increases
Solution Approach 1:
The detection system is divided into distinct functional modules: an image acquisition module that captures images, an attention network module that processes images with multi-scale feature extraction, an encoder-decoder network module that performs detailed analysis, and a multi-layer perceptron module that outputs detection results. This segmentation allows each module to specialize in specific tasks, improving overall detection accuracy while making the complex system more manageable and maintainable.
Solution Approach 2:
The attention network incorporates multi-scale feature extraction by processing images at different visual field ranges (first visual field range and second visual field range). This multi-dimensional approach allows the system to capture both fine-granularity and coarse-granularity features simultaneously, significantly improving fault detection accuracy by analyzing the same image data from multiple spatial scales.
3Measurement precision
If fine-granularity and coarse-granularity features are extracted, then measurement precision is improved, but loss of information increases
Solution Approach 1:
The encoder-decoder network merges features from multiple scales and processing stages. The encoder extracts features at different granularities, and the decoder reconstructs and fuses these features together, preserving both fine-granularity details and coarse-granularity contextual information. This merging process prevents information loss by integrating complementary features from different processing paths into a comprehensive representation for fault detection.
Data Source
AI summary
Provided is a fault detection method, an electronic device, and a storage medium. The method includes: capturing an image of a false twist component to obtain a target image; constructing a first fusion feature based on at least one attention module; for each attention module, the attention module includes a first sub-module and a second sub-module, the first sub-module constructs a first sub feature with respect to fine-granularity features within first visual field ranges and coarse-granularity features within second visual field ranges in first input information, and the second sub-module obtains a mask map; inputting the first fusion feature into an encoder network to obtain an encoded feature; decoding the encoded feature to obtain a decoded feature; obtaining a second fusion feature based on the decoded feature; and inputting the second fusion feature into a multi-layer perceptron to obtain a fault detection result of the false twist component.