基于决策树的复杂系统可靠性仿真评估

Reliability Estimation for Large Complex Systems Based on Decision Tree and M-C Simulation

提出了一种利用决策树建立经验模型进行系统可靠性的Monte-Carlo仿真评估的新方法。主要思想是通过在一个有限数据集上训练决策树得到一种计算较简单的快速评估算法,用以近似替代以往复杂费时的评估函数(EF)方法。实例计算结果表明,利用这种方法,仅需要少量的信息训练决策树就能得到对系统可靠性作出较好的估计。

A decision tree approach is presented to build an empirical machine learning model for use in Monte Carlo system rehability estima- tion.By training a decision tree on a restricted data set,an estimation algorithm is developed to replace the system performance model evalu- ation.The excellent results obtained in the examples show the potential of this method.