基于人工蜂群算法的数字电路多故障测试生成算法

A multiple faults test generation algorithm for digital circuits based on artificial bee colony algorithm

针对数字电路多固定故障测试生成故障覆盖率低和平均测试生成时间长的问题,提出了人工蜂群和神经网络优化的数字电路多固定故障测试生成算法。该算法首先通过等效的方法得到多固定故障的单固定故障模型,再使用Hopfield二值神经网络的方法得到单固定故障的约束电路模型,最后使用人工蜂群优化方法求解故障约束电路接口电路的能量函数的零值点获得原电路的多固定故障的测试生成矢量。实验结果(在ISCAS’85国际标准测试电路上)表明该算法的故障覆盖率可达98.5%以上,平均测试生成时间小于0.25μs。

A muhiple stuck-at faults test generation algorithm based on artificial bee colony and neural net- works for circuits is proposed in this paper, because the test generation faults coverage is low and the average test generation time is long for multiple stuck-at faults in digital circuits. The algorithm obtains single stuck-at fault model of multiple stuck-at faults by the method of equivalent firstly, and constructs the constraint circuit model of the single stuck-at fault circuit using Hopfield neural networks. The test vectors for multiple stuck-at faults in the original circuit can be obtained by solving the zero value of energy function of the constraint net- work＇s interface circuit with artificial bee colony optimization. The experimental results（ ISCAS＇85 internation- al standard test circuits） show the faults coverage of the algorithm is above 98.5% and the average test genera- tion time is less than 0. 25 μs.