A state assessment method for a system with non-exclusive states based on the fusion of imprecise information

Limited amount of available data is one of the major challenges when assessing the reliability of complex multi-state systems. To overcome this challenge, this paper develops an approach that employs expert judgments as an alternative over statistical data because expert knowledge is free from the requirement of collecting a large number of samples for rigorous statistical analysis. Specifically, a novel state assessment method for a system with multiple non-exclusive states is proposed based on a new uncertain information fusion framework called D number theory to fuse the imprecise information from multiple experts. In the proposed method, multiple D numbers to characterize the imprecise evaluation information of experts are constructed as reliability bounds corresponding to different acceptable states of the system. The effect of non-exclusiveness between states on the measurement and management of conflict among expert judgments is fully accounted for when carrying out fusion of multi-source imprecise information. The contributions of the work are two-fold. First, the non-exclusive degrees between non-exclusive states modeled using performance intervals are precisely quantified. Secondly, a new multi-source imprecise information fusion method for state assessment of systems with non-exclusive states is provided based on D number theory, in which the uncertainty of evaluation results is reduced notably.