基于全局最优的信息启发式Rollout测试序列生成算法

Globally Optimal Information Heuistic Rollout Test Sequence Generation Method

摘要:

为实现电子设备或电气系统快速故障诊断,在故障诊断过程中需生成测试序列,一般通过分析测试点对故障的检测与隔离的次序能够得出产品测试序列生成排故引导树。然而针对不同故障模式的故障率不同,不同测试点的测试权重、测试费用、测试时间不同,从不同测试点出发形成的排故引导树也不同。提出了一种基于A0*信息启发式Rollout策略的测试序列生成算法,其作为一种近优的结算方法既考虑可靠性,也考虑测试费用最小,既减轻庞大的计算量,又获得了比次优启发式算法更好的诊断结果。

Abstract:

In order to realize the rapid fault diagnosis of electronic devices or electrical systems, the martial arts sequence needs to be generated in the failure diagnosis. Generally, through the analysis of the martial arts test, the sequence of point-to-fault lattice and isolation can be obtained that the product test sequence generation efficiency track is a tree. However, according to the different failure rates of different failure modes, the test cane children and the cost of different test points are different, and the row guidance trees formed from different test points are also different. A test sequence generation algorithm based on A0* information heuristic Rollout strategy is proposed, which is a near-optimal settlement method that considers both reliability and minimal test cost, which not only reduces the huge amount of computation, but also obtains better diagnostic results than the suboptimal heuristic algorithm.