Research on Safety Evaluation of Social Systems from Security Viewpoints

在项目的第一阶段，将安全传感器和/或警卫的最优分配问题表述为一个图/网络问题，其中强调了多重最小割集概念的重要性。多重最小割集的概念是作者在1989年首次提出的，它是对最小割集的一种自然的、一般的推广，在图论和系统可靠性与安全性分析中都很有名。提出了求解公式化图/网络问题的算法，并从计算复杂度的角度证明了该算法是非常有效的。在项目的第二阶段，我们考虑使用基于风险的方法设计和操作安全报警系统。首先，考虑检测误差带来的风险，提出了报警系统数据项的优化设计；数据项的最优最小组合可以选择为风险最小的组合。其次，考虑到报警系统的运行，保持报警系统的可靠性是非常重要的。为了系统地确定维修时间，提出基于风险的方法：当检测错误带来的风险大于期望值时，应进行维修。该准则始终保持风险水平合理，并使不可用时间最小化。

At the first stage of the project, the problem of optimal allocation of security sensors and/or guards is formulated as a graph/network problem, where the importance of the concept of multiple minimal cut set is stressed. The concept of multiple minimal cut set, firstly introduced by the authors in 1989, is a natural and generic extension of minimal cut set which is well known in graph theory and systems reliability and safety analysis. The algorithm to solve the formulated graph/network problem is developed and it is shown that the algorithm is very effective from the view point of computational complexity.At the second stage of the project, we consider the design and operation of alarm systems for security using a risk-based approach. Firstly, the optimal design of data items for the alarm system is proposed with considering its risk caused by detection errors; the optimal minimal combination of data items can be selected as that with minimal risk. Secondly, considering the operation of an alarm system, it is very important to maintain the reliability of alarm system. To determine the maintenance time systematically, the risk-based approach is proposed as: The maintenance should be performed if the risk caused by detection errors is larger than the desired value. This criterion always keeps the risk level reasonable and minimizes the unavailable time.