用KM算法增强测试集的频谱主分量

发布时间：2018-03-01 05:17

本文关键词： 频谱分析主分量二分图匹配 KM算法相关性　出处：《电子测量与仪器学报》2017年01期 　论文类型：期刊论文

【摘要】：在集成电路测试领域常常需要对测试集和测试响应进行频谱分析,计算其频谱主分量,用于指导测试产生和进行测试数据压缩等。提出一种用KM(Kuhn-Munkras)算法增强测试集频谱主分量的方法,先根据测试集和其频谱主分量矩阵构建二分图模型和权值矩阵,把增强频谱主分量的问题转化为二分图的匹配问题,然后用KM算法求解。根据匹配关系调整测试集中测试向量的顺序后,频谱主分量和测试集的相关性增加,频谱主分量得到增强。在ISCAS-89基准电路测试集的实验表明,测试集排序后,其频谱主分量的相关性提高了19.05%,测试集残差FDR编码压缩率提高了4.59%。
[Abstract]:In the field of integrated circuit testing, it is often necessary to analyze the spectrum of the test set and test response, and to calculate the principal component of the spectrum. This paper presents a method to enhance the principal components of the spectrum of test sets by using KMN Kuhn-Munkras algorithm, which is used to guide the generation and compression of test data. Firstly, a bipartite graph model and a weight matrix are constructed according to the test set and its spectral principal component matrix. The problem of enhancing the principal component of spectrum is transformed into the matching problem of bipartite graph, and then solved by km algorithm. After adjusting the order of test vectors in test set according to the matching relation, the correlation between spectrum principal component and test set is increased. The experimental results of the ISCAS-89 reference circuit test set show that the correlation of the spectrum principal components is increased by 19.05 and the compression ratio of the test set residual FDR coding is increased by 4.59.
【作者单位】：湖南大学信息科学与工程学院;广东电网有限责任公司珠海供电局;
【基金】：国家自然科学基金(61472123)资助项目
【分类号】：TN407

【相似文献】