Neural Network Inspection Area Selection Across Product Models
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
In smart factories, determining accurate inspection areas for diverse and customized product models is challenging due to the difficulty in matching manually generated inspection areas with the actual inspection requirements, leading to incorrect or omitted inspections.
Innovation Solution
An electronic device utilizing an artificial neural network to determine inspection areas by processing input data, such as images or videos, and identifying inspection targets, while adjusting for positional relationships and excluding or adding areas based on inspection results, thereby optimizing the inspection process.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If manually generated inspection areas are used for diverse product models, then inspection coverage can be maintained, but accuracy and reliability of inspection area matching deteriorates
Solution Approach 1:
The patent replaces the manual mechanical process of generating and matching inspection areas with an automated artificial neural network system. The neural network automatically determines inspection areas based on product model information, eliminating manual intervention and improving matching accuracy while reducing operational complexity.
Solution Approach 2:
The system enables self-service by allowing the artificial neural network to autonomously generate and optimize inspection areas without human intervention. The network learns from training data and automatically adapts to different product models, making the inspection area determination process self-sufficient and highly reliable.
2Reliability
If comprehensive inspections are performed on all potential areas, then inspection completeness is improved, but inspection time and productivity deteriorate
Solution Approach 1:
The patent applies local quality by determining specific inspection areas tailored to each product model rather than uniformly inspecting all areas. The artificial neural network identifies and focuses inspection resources on relevant local regions, ensuring completeness for each model while avoiding unnecessary inspections elsewhere, thus improving productivity.
Solution Approach 2:
The inspection process is segmented into model-specific inspection areas determined by the neural network. Instead of treating all products uniformly, the system divides inspection efforts into targeted segments based on product characteristics, maintaining completeness for each segment while reducing overall inspection time through selective focus.
3Adaptability or versatility
If manual methods are used to determine inspection areas, then system simplicity is maintained, but adaptability to diverse product models deteriorates
Solution Approach 1:
The patent implements universality by designing an artificial neural network system that can handle diverse product models through a single unified platform. The neural network is trained on multiple product types and can adaptively determine inspection areas for various models, providing multi-functionality that replaces multiple manual processes while improving adaptability.
Solution Approach 2:
The system utilizes parameter changes by adjusting inspection area parameters based on product model characteristics. The artificial neural network dynamically modifies inspection parameters such as area coordinates, dimensions, and priorities according to the specific product being inspected, enabling high adaptability across diverse models through automated parameter optimization.
Data Source
AI summary
An electronic device and method for determining inspection areas of manufactured products includes obtaining first input data corresponding to a manufactured product; obtaining model information of the product; determining first candidate inspection areas by inputting the data and model information to a pre-trained artificial neural network configured to output an inspection area in response to receiving an image; performing inspections by identifying inspection targets in the candidate areas; and determining final inspection areas from among the candidate areas based on inspection results. The device may identify product models via barcodes or QR codes, perform various inspection types including fastening, shaping, and appearance inspections, and analyze positional relationships between components such as harnesses, cables, or connectors. The system adapts inspection areas by excluding non-feasible areas and incorporating newly detected areas through iterative testing and validation.