一种以模拟电路测点和指标表示最优故障判别效果的决策树实现方法

 描述了一种由测点和指标样本所表示的故障空间,给出了一种基于Fisher判别的故障分类决策过程。为了提高故障判别能力,提出了一种由测点和指标组合所表示的最优判别效果评价准则,并基于此,设计了相应的决策树算法。

A Decision Tree Implementation Approach on the Optimal Fault Discernment Effect Presented with Test Nodes and Specifications for Analog Circuit

This paper describes the fault space constructed with the samples of the test nodes and specifications, and a decision process is implemented in order to classify the fault classes based on Fisher discriminator. To improve the fault discernment, a decision tree algorithm, integrated with a novel evaluation rule, is put forward in order to gain a good discrimination effect and decrease the test cost through selecting the optimal combinations of the test nodes and specifications.