前言
创作开始时间:2021年9月28日12:25:03
基本信息
@article{van2011antares,
title={The ANTARES approach to automatic systems diagnosis},
author={van Gemund, AJC and Gupta, Shekhar and Abreu, Rui},
journal={Proceedings DX’11},
year={2011}
}
Rui Abreu 应该算定位领域的大牛了。
abstract
Recent feedback obtained when applying MBD in industry suggests that the costs involved in modeling (expertise and labor) sometimes outweighs the benefits of MBD as a high- performance diagnosis approach. In this pa- per we propose an automatic diagnosis approach, called ANTARES, that completely avoids model- ing. In this approach modeling cost is traded for increased component inspection cost as a result of the obvious sacrifice in diagnostic accuracy. We study this sacrifice for a number of logic circuits taken from the 74XXX/ISCAS benchmark suite. Our results show that ANTARES is quite capable of approaching the diagnostic performance of an MBD approach, provided accurate information is available per component on the pass rate of tests that cover the component when faulted.
- 工业界反馈:MBD(model based diagnosis)的(专家和手工)开销太大
- 所以本文trade-off一下,损失MBD精度,但是整个过程自动化。所以提出了:ANTARES 工具
- 结果表明ANTARES还不错。
introduction
In MBD the cost of the diagnostic process can be bro- ken down into modeling cost and solution cost. So- lution cost includes algorithmic cost as well as iden- tification cost (searching within the list of diagnosed candidates until the real faults are found), where iden- tification cost is often used as diagnostic utility mea- sure.
While solution cost has traditionally been an impor-
tant success factor (especially in time-critical applica- tions), a recent design and maintenance case study in Dutch industry suggests that modeling cost is much more of a bottleneck for the acceptance ofMBD than perhaps considered.
MBD度量标准的变化,挺有意思的。
Unlike MBD, in SFL the dynamic program execution profiles of tests (called spectra, hence the name SFL) is correlated with the test outcomes (pass/fail), typically by using statisti- cal similarity coefficients.
To the best of our knowledge, we are the first to pro- pose the use of SFL in the multiple-fault diagnosis of
logic circuits comprising both persistent and intermittent logic.
感觉贡献还挺大的。
小结
后面再慢慢补充。
创作结束时间:2021年9月28日13:00:00