Capturing the Information Dependencies of Verification Activities with Bayesian Networks

Capturing the Information Dependencies of Verification Activities with Bayesian Networks
复制标题

使用贝叶斯网络捕获验证活动的信息依赖性

DOI:
10.1007/978-3-030-00114-8_46
复制
发表时间:
2019
期刊:
Systems Engineering in Context
影响因子:
--
通讯作者:
Farshad Farkhondehmaal
Farshad Farkhondehmaal
中科院分区:
--
文献类型:
--
作者:
A. Salado;Hanumanthrao Kannan;Farshad Farkhondehmaal

文献摘要

被引文献

相似文献

验证活动提供了系统工程中合同履行的证据。因此,在任何采办计划中充分定义验证活动的重要性是不容置疑的。它的意义超越了合同。开发财务预算的主要部分用于执行验证活动,验证活动是发现系统知识的主要工具,这是降低开发风险的关键。因此,必须优化核查活动,以便以最少的投资实现对系统正常运作的一定程度的信任。目前优化核查战略的方法假定核查活动有助于这种置信度的绝对增加或减少。这意味着,核查活动产生的置信度与其他核查活动过去或未来可能产生的结果无关。然而,这一表述并不能准确反映实际做法。相反,是否有必要开展某一特定的核查活动,取决于先前已开展的所有核查活动的结果。为了解决这样的限制,我们在本文中展示了如何贝叶斯网络可以是一种有效的方法来捕捉验证活动的信息依赖。
Verification activities provide the evidence of contractual fulfillment in the engineering of systems. Thus, the importance of adequately defining verification activities in any acquisition program is unquestionable. Its significance extends beyond contracting though. A major portion of the development financial budget is spent in executing verification activities, and verification activities are the main vehicle in discovering knowledge about the system, which is key to reduce development risk. Hence, it is important to optimize verification activities so that a given level of confidence about the proper functioning of the system is achieved with minimum investment. Current approaches to optimize verification strategies assume that a verification activity contributes to an absolute increase or decrease in such confidence. That means that the confidence generated by a verification activity is independent of the past or potential future results of other verification activities. However, this representation is not an accurate reflection of actual practice. On the contrary, the necessity to perform a given verification activity depends on the results of all verification activities that have been previously performed. In order to address such limitation, we show in this paper how Bayesian networks can be an effective approach to capture the information dependencies of verification activities.