Similar-Defect Search System for Manufacturing Yield Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The large volume and complexity of manufacturing data make it difficult to retrieve effective information for improving productivity, as existing techniques for presenting defect frequency maps do not provide insight into the recency of defects, hindering efficient defect analysis and yield analysis.
Innovation Solution
A similar-defect search/display system that includes a determination unit, storage units for cause-unidentified and cause-identified data, and a search/display unit to identify and display manufacturing data with similar feature quantities, allowing for the classification of defect tendencies as 'recurring', 'not recently occurring', 'previously ignored', or 'newly occurring', enabling efficient defect analysis.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If defect frequency maps are presented using existing techniques, then the frequency of occurrence can be visualized, but the recency of defects cannot be determined
Solution Approach 1:
The patent segments defect data into multiple dimensions: spatial distribution (defect maps), temporal information (recency), and causal information (identified vs. unidentified causes). By dividing the comprehensive defect analysis into these separate but integrated components, the system preserves both frequency measurement and recency information simultaneously, resolving the contradiction between visualizing frequency and maintaining information completeness.
Solution Approach 2:
The patent adds a temporal dimension to the traditional defect frequency maps by introducing recency indicators that show when defects were last observed. This transforms the static frequency visualization into a dynamic multi-dimensional representation that includes time-based information, allowing users to see both how often defects occur and when they last occurred without losing either type of information.
2Productivity
If large volume of manufacturing data is collected for yield analysis, then productivity improvement potential increases, but information retrieval difficulty increases
Solution Approach 1:
The patent introduces an intermediary search and display system that sits between the large volume of collected manufacturing data and the user. This intermediary system automatically processes, indexes, and presents defect data in organized formats (defect maps with recency indicators, cause-identified vs. cause-unidentified categories), reducing the complexity of retrieving and analyzing information while maintaining access to the full data set for productivity improvement.
Solution Approach 2:
The system implements self-service capabilities by automatically organizing defect data, generating defect maps, identifying patterns, and presenting relevant information without requiring manual data processing. The automated search and display functions enable users to retrieve needed information directly from the organized data structures, reducing the complexity burden while preserving the analytical power of the collected manufacturing data.
3Ease of operation
If defect data is analyzed without recency information, then analysis simplicity is maintained, but analysis efficiency decreases
Solution Approach 1:
The patent applies preliminary action by pre-processing defect data to include recency indicators and organizational metadata before analysis is needed. Defect maps are pre-generated with temporal information, and data is pre-categorized by cause identification status. This preliminary organization maintains analysis simplicity while dramatically improving efficiency, as users can immediately see recency information without performing additional manual sorting or filtering operations.
Data Source
AI summary
According to one embodiment, system includes a determination unit, a first storage, a second storage, a search unit and a display. The determination unit determines a feature quantity of the process-targeted manufacturing data. The first storage stores cause-unidentified manufacturing data. The second storage stores cause-identified manufacturing data. The search unit searches, based on the feature quantity of the process-targeted manufacturing data, the first storage and the second storage for the cause-unidentified manufacturing data and the cause-identified manufacturing data that have a feature quantity similar to that of the process-targeted manufacturing data. The display displays the search result.