Automated Spinning Box Defect Detection with Wavelet Attention
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The manual detection of defects in spinning boxes during chemical fiber production is inefficient, reliant on experience, and prone to errors, affecting production quality and efficiency due to issues like yarn floating, breakage, misalignment, and nozzle tilt.
Innovation Solution
An automated spinning box detection method using a target detection model, such as a dynamic weights-based wavelet attention neural network (DWWA-Net), to analyze images of the spinning box for defects, including yarn floating, breakage, and nozzle tilt, providing accurate detection results on defect quantity, position, and type.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If manual detection method is used, then detection process is simple, but detection efficiency is low and accuracy is poor
Solution Approach 1:
The patent replaces the manual mechanical detection system with an automated image processing system using deep learning algorithms. The target detection model automatically identifies defects in spinning boxes by processing images, eliminating the need for manual inspection while significantly improving detection efficiency and accuracy.
Solution Approach 2:
The patent introduces an intermediary image processing system that acts as a mediator between the spinning box and the final detection result. The target detection model serves as an intermediary that processes images and provides structured defect information, bridging the gap between raw visual data and actionable inspection results.
2Measurement precision
If manual detection is used, then equipment cost is low, but detection accuracy is poor and relies on experience
Solution Approach 1:
The patent changes the parameters of the detection system by implementing a target detection model that outputs structured information including defect type, position, and size. This transformation from qualitative manual assessment to quantitative automated measurement significantly improves detection precision and consistency.
Solution Approach 2:
The patent performs preliminary action by pre-training the target detection model with labeled data before actual detection. The model learns from training images to recognize various defect patterns, enabling accurate detection without requiring manual expertise during the actual inspection process.
3Productivity
If automated detection is implemented, then detection efficiency is improved, but computational resources and time for model training are consumed
Solution Approach 1:
The patent performs the computationally intensive model training in advance as a preliminary action. Once the target detection model is trained, it can quickly process images during actual detection without requiring repeated training, thus achieving high detection speed while minimizing ongoing time consumption.
Solution Approach 2:
The patent creates a trained model copy that can be deployed for rapid detection. The trained model serves as a reusable artifact that encapsulates learned knowledge, allowing multiple detection operations to benefit from the initial training investment without repeating the time-consuming training process.
Data Source
Figure 1~2(a)
Figure 2(b)
Figure 3(a)
AI summary
The present disclosure provides a spinning box detection method and apparatus, a device and a storage medium. The method comprises: obtaining (S101) an image to be detected of a spinning box when determining that the spinning box meets a preset defect detection condition; and inputting (S102) the image to be detected into a target detection model to obtain a target detection result of the spinning box; wherein the target detection model is used to detect whether there is a defect in the spinning box to obtain the target detection result; and the target detection result comprises at least one of: a total quantity of defects, a defect position or a defect type.