Integrated circuit test data reliability evaluation method based on lightgbm algorithm

By combining the LightGBM algorithm with correlation coefficients and regression analysis, reliability labels are generated, solving the problem of identifying latent anomalies in integrated circuit testing. This enables efficient and accurate reliability evaluation, reduces retesting costs, and improves production efficiency.

CN122241096APending Publication Date: 2026-06-19HEFEI UNIV OF TECH
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
HEFEI UNIV OF TECH
Filing Date
2026-03-20
Publication Date
2026-06-19

AI Technical Summary

Technical Problem

In existing integrated circuit testing, traditional threshold judgment modes are difficult to identify implicit reliability anomalies within the acceptable range, leading to the risk of chip failure later on, and the cost of retesting is high and the efficiency is low.

Method used

The LightGBM algorithm, combined with Pearson correlation coefficient and least squares regression, is used to generate reliability labels. Latent anomalies are identified through a single test. The LightGBM algorithm is then used to process high-dimensional data to construct a reliability evaluation model.

Benefits of technology

It enables accurate identification of latent anomalies in a single test, reduces the cost of retesting, improves the comprehensiveness and accuracy of evaluation, ensures comprehensive coverage of latent anomaly data, and enhances production efficiency and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122241096A_ABST
    Figure CN122241096A_ABST
Patent Text Reader

Abstract

This invention provides a reliability evaluation method for integrated circuit test data based on the LightGBM algorithm, relating to the field of integrated circuit testing. This invention combines precise theoretical calculations with machine learning algorithms, establishing a method for evaluating the reliability of integrated circuit test data through data preprocessing, correlation calculation, data reliability label generation, and model training. This method can effectively identify abnormal data in integrated circuit test data, generate reliability evaluation labels, and accurately classify different types of test data. Simultaneously, this method can provide reliability evaluation of integrated circuit test data, filter unreliable data during the testing process, improve the credibility of test results, and provide technical support for subsequent fault diagnosis and failure analysis.
Need to check novelty before this filing date? Find Prior Art