A New Uncertainty Importance Measure in Fuzzy Reliability Analysis

Uncertainty is inevitable in any reliability analysis of complex engineering systems due to uncertainties present in models, parameters of the model, phenomena and assumptions. Uncertainties at the component level are propagated to quantify uncertainty at the system level reliability. It is very important to identify all the uncertainties and treat them effectively to make reliability studies more useful for decision making. Conventional probabilistic approaches adopt probability distributions to characterize uncertainty where as fuzzy reliability models adopt membership functions to characterize uncertainty. Both the approaches are widely used in uncertainty propagation for reliability studies. However, identification of critical parameters based on their uncertainty contribution at the system level is very important for effective management of uncertainty. A method is proposed here in the fuzzy framework to rank the components based on their uncertainty contribution to the over all uncertainty of system reliability. It is compared with probabilistic methods using a practical reliability problem in the literature.