Abstract：With the constant improvement of the scale and complexity of integrated circuit,how to efficiently generate test vector is becomed the key to digital circuit board fault detection. Based on the analysis of the automatic test vector generation questions, the mathematical model of circuit under test by using neural network is built, which convert the test vector generation problems into math problems, and puts forward a kind of efficient particle swarm optimization algorithm for solving the problems. VC++ programming tool is used to realize the proposed approach.Some of the international standard circuits ISCAS '85 are carried on the experiment.The experiment data shows that the fault coverage is up to 100%,the test time for small-scale single-fault circuit is reduced by 13% compared with reference, the large-scale circuit test time reduction of 61%, and the larger the circuit scale and reduce the time more obvious.
杨慎涛,刘文波. 基于粒子群的神经网络测试生成算法[J]. 计量学报, 2015, 36(2): 197-201.
YANG Shen-tao, LIU Wen-bo. Test Generation Algorithm of Neural Networks Based on the PSO. Acta Metrologica Sinica, 2015, 36(2): 197-201.
［1］许杰,涂小辉.基于改进遗产算法的组合电路测试生成［J］.实验室研究与探索,2012,31(7):76-78.
［2］Fujiwara H. Three-valued neural networks for test generation［C］//Proc of 20 the Fault-Tolerant Computing Symposium,Newcastle Upon Tyne, UK,1990.
［3］吴丽华.数字集成电路多故障测试生成算法和可测性设计的研究［D］.哈尔滨:哈尔滨理工大学,2007.
［4］郭希维,苏群星,谷宏强.数字电路测试中的关键技术研究［J］.科学技术与工程,2006,18(6):2903-2905.
［5］张琪,侯加林,闫银发,等.基于虚拟仪器的电路板故障检测与诊断系统的研究［J］.电子测量与仪器学报,2011,25（2）:135-140.
2015, Vol. 36 
