论文部分内容阅读
本文研究将人工神经网络用于组合电路测试产生的一般模型,分析影响这一方法效率的因素,提出了用于降低被测电路对应网络规模的故障压缩,电路分块,多级蕴涵等策略,采用改进的梯度算法加速了网络能量函数极小值的搜索。介绍了基于这些策略开发的一个测试生成系统的结构。实验结果说明了提出方法的有效性。
In this paper, artificial neural network (ANN) is applied to the general model of combinational circuit test. The factors influencing the efficiency of this method are analyzed. Some strategies such as fault compression, circuit block and multi-level implication are proposed to reduce the network scale of the circuit under test. An improved gradient algorithm is used to accelerate the search of the minimum value of the network energy function. The structure of a test generation system based on these strategies is introduced. The experimental results show the effectiveness of the proposed method.