Quality Defect Diagnosis Using Segmented Data Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing quality prediction models fail to accurately predict and diagnose quality defects due to the lack of consideration for different mechanisms causing various types of defects, leading to insufficient prediction performance.
Innovation Solution
A quality defect diagnosis device and method that divides product data into predetermined blocks, aggregates operational data based on defect trends, and uses multiple prediction models to classify and estimate defects, allowing for high-accuracy prediction and defect factor estimation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If quality data are used uniformly regardless of their characteristics in predicting product quality, then the prediction model can be generated with simple data processing, but the prediction performance is insufficient due to lack of information about different types of quality defects caused by different mechanisms
Solution Approach 1:
The patent segments quality data into multiple classification groups based on defect characteristics and causes. The data classification unit divides quality data into different categories (e.g., surface defects, internal defects, dimensional defects) and the data aggregation unit creates separate aggregated data for each classification. This segmentation allows the prediction model to learn specific patterns for each defect type, improving prediction accuracy while maintaining manageable complexity through systematic organization.
2Reliability
If multiple prediction models are generated for each classification of quality defects, then high-accuracy prediction and defect factor estimation are achieved, but the system complexity increases
Solution Approach 1:
The patent implements a universal data processing framework that handles multiple defect classifications through a single integrated system. The data aggregation unit performs unified aggregation operations (calculating means, maximums, minimums, standard deviations) for all defect types, and the prediction model generation unit uses consistent machine learning procedures across all classifications. This multi-functional approach achieves high prediction accuracy for various defect types while avoiding the complexity of completely separate processing systems for each defect category.
Data Source
Figure 1
Figure 2
Figure 3
AI summary
A quality defect diagnosis device diagnoses a quality defect occurring in a product, the quality defect diagnosis device including: a data dividing unit that two-dimensionally divides product data as a diagnosis subject into predetermined blocks, the product data including operational data and quality data; a data aggregation unit that extracts the operational data for each of predetermined classifications of quality defects on a basis of a trend of quality defects in the divided blocks and aggregates the extracted operational data for each of the blocks and for each region including two or more of the blocks; a quality defect determination unit that evaluates presence or absence of occurrence of a quality defect by using a plurality of prediction models which is generated for each of the classifications of quality defects and in which a relationship between the aggregated operational data and the quality data is learned; and a factor estimation unit that estimates a factor of the quality defect on a basis of a quality defect quality defect determination step result.