神经网络和优化算法在数字系统测试中的应用  

Application of Neural Network and Optimization Algorithm in the Test of Digital System

作　　者：陆广平[1] 王友仁[1] 

机构地区：[1]南京航空航天大学自动化学院,江苏南京210016

出　　处：《计算机测量与控制》2006年第2期164-165,174,共3页Computer Measurement &Control

摘　　要：介绍了用Hopfield神经网络模型把组合电路测试转化为相应的能量函数,采用遗传算法、最速下降法结合模拟退火法的优化算法来求解给定故障的测试矢量;通过并行故障仿真检验测试矢量集的故障覆盖率,并压缩测试矢量集,然后将电路响应序列Q移入特征多项式求得特征R,由此建立故障字典。Combinational circuits are converted into energy functions with Hopfield neural network models, optimization algorithm of genetic algorithm and fast descent annealing algorithm are used to obtain test vectors of stuck faults. Fault coverage of test vectors are detected through parallel fault simulation, test vectors are compacted, then character sequences are obtained through shifting response sequences to character multinomial, in conclusion fault dictionary is built.

关 键 词：神经网络 优化算法 测试矢量 特征序列 故障字典 

分 类 号：TP306[自动化与计算机技术—计算机系统结构]