基于证据理论的异类信息统一表示与建模

Unified Method of Describing and Modeling Heterogeneous Information Based on Evidence Theory

在复杂系统的性能、可靠性评估中,由于信息不完备、实验样本有限、对复杂物理过程的认识不充分等原因,存在多种不确定性。由于不确定性的来源不同,因此在系统建模时不确定性参数存在概率分布、模糊分布、概率包络、区间信息等多种表达方式,难以用统一的方法对系统的性能、可靠性进行量化评估。研究了概率分布、概率包络、模糊分布、专家估计信息、小样本测试数据等不同类型的信息转化为基于证据理论的统一表示的方法,讨论了基于证据理论的异类信息统一表示方法在模型不确定性量化中的应用。实例仿真表明了所提出方法的有效性。

In reliability analysis and assessment of complex system, it is always in the situation that experimental data is limited, or information is not complete, which produces a variety of uncertainties in the system. As sources of uncertainties are different, there are some different forms of uncertainty representations in system modeling, such as probability distribution, fuzzy distribution, probability box, small sample information, and so on. To solve the problem that system performance and reliability are hard to be assessed with heterogeneous information, the method of translating heterogeneous information into evidence body uniformly was studied. Then the application of proposed method in model uncertainty auantitTcation was described. Simulation results show the effectiveness oflaroposed method.