Adhesive Bond Coverage Scoring for Vehicle Component Inspection
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Manual assessment of adhesive bond quality on vehicle components is subjective and time-consuming, leading to inconsistencies in evaluating the coverage of adhesive bonding, which affects the performance of bonded components.
Innovation Solution
A neural network-based system that generates surface features, classifies inadequate coverage regions, and predicts a coverage score using image segmentation techniques, applying rules to generate a bond quality assessment.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If manual assessment of adhesive bond quality is used, then flexibility and adaptability to different components are maintained, but subjectivity and time consumption increase leading to inconsistencies
Solution Approach 1:
The patent replaces the manual mechanical assessment process with an automated neural network-based image analysis system. The neural network processes images of adhesive bonds to automatically detect and classify coverage regions, eliminating human subjectivity and significantly reducing assessment time while maintaining high precision through consistent application of the same evaluation criteria across all components
Solution Approach 2:
The system creates a digital copy (image) of the physical adhesive bond and processes this copy through the neural network for assessment. This allows the actual component to remain undisturbed while enabling rapid, repeated analysis of the bond quality without physical contact or manipulation of the original part
2Measurement precision
If manual assessment of adhesive bond quality is used, then complex judgment criteria can be applied, but subjectivity increases leading to inconsistent evaluations
Solution Approach 1:
Complex manual judgment criteria are replaced with a trained neural network that automatically applies consistent evaluation rules. The neural network has been trained on labeled data to recognize adequate and inadequate coverage patterns, eliminating variability in human judgment while maintaining the ability to handle complex bond configurations through learned patterns rather than explicit programming
Solution Approach 2:
The neural network is pre-trained on a dataset of labeled adhesive bond images before deployment. This preliminary training phase allows the system to learn complex evaluation criteria from examples, so that when actual assessment is performed, the complex judgment logic is already embedded in the trained model, simplifying the runtime operation while maintaining high evaluation accuracy
3Productivity
If automated neural network assessment is implemented, then assessment speed and consistency improve, but system complexity and implementation difficulty increase
Solution Approach 1:
The neural network system is designed to be universal and adaptable to different adhesive bond types and components. Rather than requiring separate systems for each application, the same neural network framework can be trained on different datasets to assess various bond configurations, reducing overall system complexity while maintaining high productivity across diverse manufacturing scenarios
Solution Approach 2:
The neural network system is designed to be self-sufficient in its assessment operations. Once trained, it autonomously processes images, generates predictions, and outputs results without requiring constant human intervention or complex external control systems. The system handles the entire assessment workflow independently, from image processing to quality determination, simplifying implementation while maximizing throughput
Data Source
AI summary
Systems and methods for bond quality assessment are provided. In one example, operations include receiving an input image of a vehicle component have adhesive bonding at a surface of the vehicle component. The operations cause a neural network to generate a set surface of features based on a set of training images. The operations also cause a neural network to classify one or more of the surface features as one or more inadequate coverage regions of the adhesive bonding. The operations further cause the neural network to generate a mapping. The operations cause the neural network to predict a coverage score for the vehicle component based on the one or more inadequate coverage regions. The operations include applying a set of rules to the mapping and the coverage score to generate a bond quality assessment of the vehicle component. The operations include executing a vehicle component fabrication plan.