电路测试神经网络方法中求多个测试矢量  

Generating Multiple Test Patterns in Digital Circuits Test Algorithm Based on Neural Networks

作　　者：潘中良[1] 张光昭[2] 

机构地区：[1]华南师范大学物理系,广州510631 [2]中山大学电子系,广州510275

出　　处：《微电子学与计算机》2000年第3期25-28,共4页Microelectronics & Computer

摘　　要：文章研究在数字电路测试的神经网络方法中求给定故障对应的多个测试矢量的方法 ,首先提出了一种求多个测试矢量的遗传进化方法 ,然后提出了一种矢量扰动方法 ,通过这两者的结合使用 ,能获得被测电路较小的完备测试集 ,从而提高了电路测试神经网络方法的性能。The methods to generate multiple test patterns in the test generation algorithms based on artificial neural networks are studied. A genetic algorithm is presented, which is able to generate several tests for a fault in the circuit under testif the fault have more than one test pattern. The scale of test set is reduced, and the performances of test generation based on neural networks are improved.

关 键 词：数字电路 神经网络 进化算法 电路测试 测试矢量 

分 类 号：TN790.7[电子电信—电路与系统]