证据理论合成准则的一种新算法及其验证  

A New Algorithm for Evidence Combination Rule and Its Validation

作　　者：冷宣兵[1] 王平[1] 张立[1] 

机构地区：[1]海军工程大学管理工程系,湖北武汉430033

出　　处：《计算机仿真》2010年第2期162-165,共4页Computer Simulation

摘　　要：研究证据理论是一种处理不确定问题的重要方法,以往的改进方法大都只关注证据的冲突问题。针对证据合成准则在不确定的证据源较多和识别的框架较大时计算复杂,容易得到错误合成结果,根据出现的问题,提出了一种证据合成准则的简便计算公式,公式模块化程度高,易于编程实现,降低了计算难度。给出了公式的具体算法步骤,算法主要通过矩阵运算来完成。仿真实例表明相对于原始公式,计算简便,易于实现,同时仿真结果也验证了公式和算法的正确性与有效性。Evidence theory is an important method to deal with uncertainty problem. The former improvements almost focus on the evidence contradiction. The calculation is complicated and the combination conclusion is prone to be wrong when there are many uncertain evidence sources and the discrimination frame is large. A simple evidence combination formula is proposed to overcome the problem. This formula has a high degree of modularization and is easy to be achieved by program, which reduces the difficulty of calculation. The detailed algorithm process is given and it is achieved by matrix arithmetic. The simulation example shows that this formula is simpler and is easier to be realized than the primal. The simulation result also proves the correctness and effectiveness of the formula and algorithm.

关 键 词：证据理论 合成准则 算法 矩阵 

分 类 号：TP301[自动化与计算机技术—计算机系统结构]