In the analysis of the risk associated to rare events that may lead to catastrophic consequences with large uncertainty, it is questionable that the knowledge and information available for the analysis can be reflected properly by probabilities. Approaches other than purely probabilistic have been suggested, for example using interval probabilities, possibilistic measures, or qualitative methods. In the present paper, we look into the problem and identify a number of issues which are foundational for its treatment. The foundational issues addressed reflect on the position that "probability is perfect" and take into open consideration the need for an extended framework for risk assessment that reflects the separation that practically exists between analyst and decision maker.
In risk analysis, the treatment of the epistemic uncertainty associated to the probability of occurrence of an event is fundamental. Traditionally, probabilistic distributions have been used to characterize the epistemic uncertainty due to imprecise knowledge of the parameters in risk models. On the other hand, it has been argued that in certain instances such uncertainty may be best accounted for by fuzzy or possibilistic distributions. This seems the case in particular for parameters for which the information available is scarce and of qualitative nature. In practice, it is to be expected that a risk model contains some parameters affected by uncertainties that may be best represented by probability distributions and some other parameters that may be more properly described in terms of fuzzy or possibilistic distributions. In this article, a hybrid method that jointly propagates probabilistic and possibilistic uncertainties is considered and compared with pure probabilistic and pure fuzzy methods for uncertainty propagation. The analyses are carried out on a case study concerning the uncertainties in the probabilities of occurrence of accident sequences in an event tree analysis of a nuclear power plant.
International audienceWe look at different prognostic approaches and the way of quantifying confidence in equipment Remaining Useful Life (RUL) prediction. More specifically, we consider: (1) a particle filtering scheme, based on a physics-based model of the degradation process; (2) a bootstrapped ensemble of empirical models trained on a set of degradation observations measured on equipments similar to the one of interest; (3) a bootstrapped ensemble of empirical models trained on a sequence of past degradation observations from the equipment of interest only. The ability of these three approaches in providing measures of confidence for the RUL predictions is evaluated in the context of a simulated case study of interest in the nuclear power generation industry and concerning turbine blades affected by developing creeps. The main contribution of the work is the critical investigation of the capabilities of different prognostic approaches to deal with various sources of uncertainty in the RUL prediction
Equipment condition monitoring of nuclear power plants requires to optimally group the usually very large number of signals and to develop for each identified group a separate condition monitoring model. In this paper we propose an approach to optimally group the signals. We use a Genetic Algorithm (GA) for the optimization of the groups; the decision variables of the optimization problem relate to the composition of the groups (i.e., which signals they contain) and the objective function (fitness) driving the search for the optimal grouping is constructed in terms of quantitative indicators of the performances of the condition monitoring models themselves: in this sense, the GA search engine is a wrapper around the condition monitoring models. A real case study is considered, concerning the condition monitoring of the Reactor Coolant Pump (RCP) of a Pressurized Water Reactor (PWR). The optimization results are evaluated with respect to the accuracy and robustness of the monitored signals estimates. The condition monitoring models built on the groups found by the proposed approach outperform the model which uses all available signals, whereas they perform similarly to the models built on groups based on signal correlation. However, these latter do not guarantee the robustness of the reconstruction in case of abnormal conditions and require to a priori fix characteristics of the groups, such as the desired minimum correlation value in a group.
Prognostic models should properly take into account the effects of operating conditions on the degradation process and on the signal measurements used for monitoring. In this paper, we develop a particle filter-based (PF) prognostic model for the estimation of the remaining useful life (RUL) of aluminum electrolytic capacitors used in electrical automotive drives, whose operation is characterized by continuously varying conditions. The capacitor degradation process, which remarkably depends on the temperature experienced by the component, is typically monitored by observing the capacitor equivalent series resistance (ESR). However, the ESR measurement is influenced by the temperature at which the measurement is performed, which changes depending on the operating conditions. To address this problem, we introduce a novel degradation indicator independent from the measurement temperature. Such indicator can, then, be used for the prediction of the capacitor degradation and its RUL. For this, we develop a particle filter prognostic model, whose performance is verified on data collected in simulated and experimental degradation tests
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.