基于神经网络的潜在通路分析

针对传统潜在通路分析过程复杂、劳动量大的缺陷，提出了基于神经网络的潜在通路分析。神经网络的输入为系统元件的定性状态组合，输出为预测实现功能组合。利用元件与设计功能之间的关系，形成神经网络训练样本。经过训练后，神经网络预测所有元件状态组合实现的功能，然后通过与设计功能的比较，确定电路的潜在通路。经仿真验证，此方法进行潜在通路分析时的正确率达到了92％，而且分析的工作量比较小。

Sneak Circuit Analysis Based on Neural Network

The paper presented a novel approach of sneak circuit analysis based on neural network, which inputs were the vectors of qualitative states of components of system and outputs were the vectors of designedly functional combination. The novel approach can overcome the weaknesses of conventional methods of sneak circuit analysis, such as the complexity of analysis and laborious analysis process. The relations between components of circuit and designed functions generated the samples which were used to train neural network, the trained neural network was used to predict the functions of circuit with a given switch state vector. The sneak circuit can be discovered by comparing the predicted functions designed ones. The simulation revealed that the legitimacy of the approach is more than 92 percent, and the workloads were reduced greatly.