Chip Test Data Detection Using Distribution Pattern Analysis
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current chip testing methods fail to accurately detect abnormal chips within the normal distribution range, leading to potential quality issues and accidents in mass production.
Innovation Solution
A detection method that analyzes historical test data to establish target test limit values based on distribution patterns, allowing for real-time detection of abnormal data and dynamic adjustment of test limits to ensure chip quality.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional test specification range detection is used, then detection simplicity is maintained, but detection precision deteriorates due to inability to identify abnormal chips within normal range
Solution Approach 1:
The patent segments the detection process into multiple stages: first determining whether test data falls within the test specification range, and second determining whether the data follows a normal distribution pattern. This segmentation allows the system to maintain simplicity for clearly passing/failing cases while adding statistical analysis only when needed, thus improving detection precision without proportionally increasing overall complexity.
Solution Approach 2:
The patent performs preliminary filtering by first checking if test data falls within the specification range before conducting more complex normal distribution analysis. This preliminary action eliminates obviously passing or failing cases early, reducing the computational burden of subsequent statistical tests and maintaining efficiency while improving overall detection accuracy.
2Measurement precision
If dynamic test limit adjustment is implemented, then detection accuracy is improved, but processing time increases due to historical data analysis
Solution Approach 1:
The patent performs preliminary determination of whether test data is within specification range before conducting time-consuming normal distribution analysis on historical data. This preliminary filtering reduces the volume of data requiring complex statistical processing, thereby improving test limit accuracy while minimizing the time penalty associated with historical data analysis.
Solution Approach 2:
The patent applies normal distribution analysis selectively rather than universally - only to test data that falls within the specification range. This partial application of the more time-consuming analytical method achieves improved detection accuracy for borderline cases without incurring the full time cost of analyzing every data point, thus balancing accuracy improvement with processing time constraints.
Data Source
AI summary
The present invention discloses a detection method, a system, an electronic equipment, and a storage medium of product test data, where the detection method includes: obtaining historical test data of historical batches of products; screening the historical test data to obtain intermediate test data; grouping the intermediate test data based on preset test parameters to obtain first groups; obtaining distribution patterns of the first groups based on the intermediate test data of the first groups; when the distribution pattern is a preset distribution pattern, using the first group corresponding to the distribution pattern as a target group; and obtaining a target test limit value based on the intermediate test data corresponding to the target group. In the present invention, the test limit value can be adjusted dynamically and adaptively, and chip test data with abnormal data can be effectively detected in real time, which improves test quality of the chip.