Neural Network Coating Defect Analysis Apparatus
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current methods for analyzing coating defects, particularly coating irregularities on surfaces, are time-consuming and labor-intensive, requiring extensive input of data before determining the cause of the defect.
Innovation Solution
An analysis apparatus and program utilizing a learned neural network that processes image data from abraded coating surfaces to quickly and accurately identify the cause of coating irregularities, with options for adjusting abrasion levels and selecting neural networks based on coating color for enhanced accuracy.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If coating defect data and coating condition data are input to determine the cause of coating defects, then the cause analysis can be performed, but the process becomes troublesome and time-consuming
Solution Approach 1:
The patent replaces the manual data input and analysis process with an automated image processing system. A camera captures images of coating defects, and image processing automatically extracts defect characteristics, eliminating the need for manual data entry and significantly reducing analysis time while maintaining diagnostic accuracy
Solution Approach 2:
The patent creates a visual copy of the coating defect through image capture. Instead of manually collecting and inputting various defect parameters, the system captures a complete visual representation of the defect, which is then processed to automatically determine the cause, thereby simplifying the analysis process
2Measurement precision
If the surface layer of coating layer is abraded to expose dust for accurate determination, then the determination accuracy is improved, but the preparation work becomes more complex
Solution Approach 1:
The patent performs preliminary abrasion of the coating surface layer before image capture to expose underlying dust particles. This preparatory action ensures that the defect characteristics are clearly visible in the captured image, improving determination accuracy while the automated image processing subsequently handles the analysis
3Measurement precision
If multiple learned neural networks are prepared for respective coating colors, then the determination accuracy is improved, but the system complexity increases
Solution Approach 1:
The patent segments the neural network system according to coating colors. Multiple learned neural networks are prepared, each specialized for a specific coating color. The system automatically selects the appropriate neural network based on the detected coating color, thereby improving determination accuracy for different color types while managing system complexity through organized modularity
Data Source
AI summary
An analysis apparatus includes: a storage unit configured to store a learned neural network; an acquisition unit configured to acquire determination image data obtained by capturing an image of a surface in which at least a surface layer of a coating layer stacked on the coating irregularity generated in the coating surface to be determined has been abraded; a determination unit configured to input the determination image data acquired by the acquisition unit to the learned neural network read out from the storage unit and determine the cause of occurrence of the coating irregularity generated in the coating surface to be determined; and an output unit configured to output information regarding the cause of occurrence determined by the determination unit.