用决策树方法优化表决器

通过分析多数表决器的决策树,并根据决策树的分裂规则,对多数表决器的决策树进行扩展。扩展方法是在决策树的一个叶子节点上进行分裂,该叶子节点为多数表决器处于无法表决的环节。分裂属性是针对是否使用加权表决提出的阈值,该阈值用于选择性地使用加权表决算法进行表决。实际测试表明所改进的多数加权混合表决器在三模冗余系统中发生二模块错误或三模块错误时,能够提高表决正确率,降低无法表决情况的概率。

Applying Decision Tree to Optimize Voter

Approaching optimize majority to decision tree,then complying split rule,then extending the decision tree of majority voter.The strategy is split one sub node.This sub node represents the situation of benign vote result.Adding a threshold to voting procedure to determine using weight value algorithm or not.In double or triple error injection test,majority-value voter showed higher ratio of correct outputs than majority voter.Furthermore,it has a lower ratio of benign signals.