基于Hopfield神经网络数字系统测试技术研究

介绍了用离散Hopfield神经网络模型把组合电路约束网络转化为能量函数，用数学优化求能量函数的最小值，即为给定固定型故障的测试矢量。经检测故障覆盖率达到100％并通过试探法进一步优化测试矢量集，然后将测试矢量集的响应序列移入本原多项式求得特征序列，建立故障字典，实验证明该方法切实有效。

Study of Digital System Testing Technology Based on Hopfield Neural Network Models

In the present research, the authors converted restricted networks of combinational circuits into energy functions with discrete Hopfield neural network models, used ptimization algorithm to obtain minimum of energy functions,the test vectors of stuck faults and found fault coverage is 100 percent. They optimized test vectors with putting out feelers, obtained character sequences of test vectors through shifting response sequences to fountain multinomial, and builded up a fault dictionary , whose validity stood up to experimental results.