Evaluation And Optimal Design Of Security & Protection Systems Based On Evidence Theory

Security & protection systems have become ubiquitous in our lives. However,due to the lack of effective risk assessment methods for security systems and the corresponding optimization design methods, currently most of the security systems in China are only simple combinations and subjective layouts of security products and prone to over-protection or inadequate protection in key areas and other issues. The establishment of a comprehensive, quantitative assessment and optimization design method for security & protection systems is a central issue of China’s security field now.The evidence theory is applied to the risk assessment and optimization design process of security & protection systems in this paper. The detailed process of the proposed method is illustrated through a example model. The main innovative research results are as follows:(1) The evidence theory is first systematically applied to the risk assessment process of security protection systems. A complete risk assessment method based on information fusion is proposed. The two-dimensional planar modeling for security system is carried out, which can characterize the system structural information more effectively; The final risk information can be obtained through combination of multiple component risk information. This final risk information can effectively express the comprehensive protection of the plane point; Also, the assessment result can be three-dimensionally visualized, which is more intuitive.(2) The plane point’s risk information is first expressed by the basic probability assignment(BPA), which can handle both the uncertain and unknown situations and reduce the information loss. Three BPA generation models are presented, namely the BPA generation model based on normal distribution, the non-parametric BPA generation model based on Gaussian process regression, and the BPA generation model based on fuzzy membership. Especially, the third model provides a way for the system staff to customize strategies on threat generation. Threat is then transformed into BPA with the help of fuzzy membership functions.(3) In order to achieve the probabilistic decision-making of risk assessment, two methods are proposed to transform BPA to probability in this paper, namely the evidence correlation coefficient-based transformation, and the conflict coefficient-based transformation. By comparing them with the classical pignistic probability transformation method in several examples, the recommendation to use these three methods is given.(4) For the optimization design problem of security system, a method to construct the mathematical model of optimization is proposed, and the solving methodology for this problem is presented. By the combination of DIRECT algorithm and genetic algorithm, each method’s advantage is realized. The optimization design can help to achieve best allocation of components under fixed resource conditions, thus maximizes the effectiveness of system protection, which will help improve the system’s cost-performance ratio.