安全防范系统的风险评估

发布时间：2018-06-13 08:31

本文选题：关键基础设施 + 安全防范系统　；参考：《西南大学》2017年硕士论文

【摘要】：关键基础设施作为一个国家的根本。关键基础设施是指具有任何在国家安全与运转体系中扮演重要角色的系统或者设施。它们不仅包括电力以及水处理工厂等设施,而且还涵盖了比如化工、金融服务、信息技术等相当广泛的行业与设施范畴。近年来,由于国际社会的持续动荡,特别是像恐怖主义这样的人为因素,对关键基础设施的安防系统构成了巨大的威胁。这迫使我们需要进一步重视相关安全防范系统的维护与建设。面对一个大型、复杂的安全防范系统,我们如何评估它的有效性;如何判断系统中是否存在漏洞、哪些地方更加关键等等。这都需要我们提出一种合理、客观的方法对一个安全防范系统进行建模,分析它的有效性、脆弱性以及所面临的风险。针对以上问题,本文在失效模式及其影响分析模型的基础上,构建了一个评估安全防范系统风险的评估框架。本文的的主要流程是首先利用改进的多目标敌对者序列图将一个关键基础设施的平面图转化为一个安防网络。再利用风险熵方法计算网络中每个节点的防范效能,然后将节点上的安防效能转化为安防网络中边的权重。接着用改进的变形虫算法计算出到达每个目标的最短路径,这些最短路径就是到达一个目标的所有路径中防范效能最弱的路径。最后我们利用失效模式及其影响分析模型来计算每条路径的风险值,最终确定系统中面临风险值最高的目标和路径。在这个评估流程中,本文的主要贡献有三点:(1)在评估安防节点的防护效能部分,不同于多数文献使用AHP方法来确定每个指标的权重,本文提出了有序可视图加权平均算子来得到每个指标的权重。该方法克服了AHP方法需要专家凭经验主观输入权重的缺点,并且解决了当专家对指标间的相对权重一无所知,无法使用AHP方法得到权重的情况。相对于经典的有序加权平均算子仅考虑输入数据序的信息,本方法还考虑了数据值本身的信息,这使得本方法更加合理。相对于AHP方法需要专家输入每个指标的相对权重,本方法是数据驱动、无参输入的,加权方式更加客观。此外,文中还用了一个应用和一个例子来说明了本方法的有效性。(2)在将设施平面图转化为安防网络部分,本文改进了原始的仅针对单个目标的敌对者序列图。在现实情况中,一个设施内往往有多个目标需要被保护。并且,当系统探测到一个入侵者时,有时候我们并不知道他的目标具体是哪一个。因此,我们有必要将只针对一个目标分析的敌对者序列图拓展到多个目标。同时,在生成安防网络边上的权重的时候,不同于原始的敌对者序列图将两个区域间的距离设为默认值,改进的多目标序列图引入了条件因素。这使管理人员可以加入像路径隐蔽性等这类其他因素,使模型的建立更加灵活。此外,因为系统中有多个目标,所以我们改进了原始的变形虫算法,使之能解决单源多汇的问题,避免了多次运行最短路径算法求解每个目标最薄弱路径的繁琐和低效率。(3)在计算每条路径面临的风险值部分,本文提出了一种将专家的定性评估结果转化为定量结果的方法。该方法的主要步骤是首先用三角模糊数将此类定性的语言变量量化。再用三角模糊数的距离公式,将多个专家的评估结果聚合,得到每个目标以三角模糊数形式表示的综合评价值。最后利用逼近最优解的偏好顺序法得到每个模式失效的后果严重程度的排序值,最终得到了一个清晰的值。该方法的主要优点是比已有的方法更加简单。并且,本文还用了一个例子来说明了本方法的有效性。
[Abstract]:Key infrastructure is the root of a country. Key infrastructure refers to any system or facilities that play an important role in national security and operation systems. They include not only electricity and water treatment plants, but also a wide range of industries and facilities such as chemical, financial services, and information technology. In recent years, the continuing turbulence in the international community, especially the human factors such as terrorism, poses a great threat to the security system of key infrastructure. This has forced us to pay more attention to the maintenance and construction of the related security system. How to estimate its effectiveness, how to judge whether there are loopholes in the system, which is more critical, and so on. It all requires us to put forward a reasonable and objective method to model a security system and analyze its effectiveness, vulnerability and risk. In this paper, the failure mode and its impact analysis model are presented in this paper. The main flow of this paper is to transform the plane map of a key infrastructure into a security network first, and then use the risk entropy method to calculate the prevention efficiency of each node in the network. The security effectiveness of the point is transformed into the weight of the side in the security network. Then the shortest path to each target is calculated with the improved amoeba algorithm. These shortest paths are the weakest paths in all paths to a target. Finally, we use the failure mode and its impact analysis model to calculate each path. In this evaluation process, the main contribution of this paper is three points: (1) in evaluating the protection effectiveness part of the security node, different from the majority of the documents using the AHP method to determine the weight of each index, this paper proposes an ordered view weighted mean operator. To the weight of each index, the method overcomes the disadvantage that the AHP method requires the expert to enter the weight with the experience subjective input weight, and solves the situation that the expert can not use the AHP method to get the weight when the expert knows nothing about the relative weight between the indexes. Considering the information of the data value itself, this makes the method more reasonable. Relative to the AHP method, the relative weight of each index is required by experts. This method is data driven, no input, and the weighting method is more objective. In addition, an application and an example are used to illustrate the effectiveness of the method. (2) in the facility plan As part of the security network section, this article improves the original sequence diagram of an opponent for a single target. In reality, there are often multiple targets in a facility that need to be protected. And, when the system detects an intruder, sometimes we do not know what the target is. The sequence diagram of a target analysis is extended to multiple targets. At the same time, when the weight of the security network side is generated, the distance between the two regions is set as the default, and the improved multi target sequence diagram introduces the condition factors. This makes the manager join the path concealment. Other factors make the model more flexible. In addition, because there are multiple targets in the system, we have improved the original amoeba algorithm to solve the problem of single source and multiple sinks and avoid the complexity and low efficiency of the shortest path algorithm to solve the weakest path of each target. (3) calculating the wind faced by each path. The main step of this method is to quantify the qualitative language variables by trigonometric fuzzy numbers, and then use the distance formula of triangular fuzzy numbers to aggregate the evaluation results of several experts, and get each target in the form of triangular fuzzy numbers. Finally, a clear value is obtained by using the preference order method of approximating the optimal solution to the seriousness of the severity of the consequences of each pattern. The main advantage of this method is more simple than the existing method. Furthermore, a case is used to illustrate the effectiveness of the method.
【学位授予单位】：西南大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP301.6;D631.4

【相似文献】