Irina Crenguta Popescu

FROM FUZZY CLUSTERING TO A FUZZY RULE-BASED FAULT CLASSIFICATION MODEL

The article overviews current trends in research studies related to reliability prediction and prognostics. The trends are organized into three major types of prognostic models: failure data models, stressor models, and degradation models. Methods in each of these categories are presented and examples are given. Additionally, three particular computational prognostic approaches are considered; these are Markov chain-based models, general path models, and shock models. A Bayesian technique is then presented which integrates the prognostic types by incorporate prior reliability knowledge into the prognostic models. Finally, the article also discusses the usage of diagnostic/prognostic predictions for optimal control.

A fuzzy decision tree for fault classification.

In plant accident management, the control room operators are required to identify the causes of the accident, based on the different patterns of evolution of the monitored process variables thereby developing. This task is often quite challenging, given the large number of process parameters monitored and the intense emotional states under which it is performed. To aid the operators, various techniques of fault classification have been engineered. An important requirement for their practical application is the physical interpretability of the relationships among the process variables underpinning the fault classification. In this view, the present work propounds a fuzzy approach to fault classification, which relies on fuzzy if-then rules inferred from the clustering of available preclassified signal data, which are then organized in a logical and transparent decision tree structure. The advantages offered by the proposed approach are precisely that a transparent fault classification model is mined out of the signal data and that the underlying physical relationships among the process variables are easily interpretable as linguistic if-then rules that can be explicitly visualized in the decision tree structure. The approach is applied to a case study regarding the classification of simulated faults in the feedwater system of a boiling water reactor.

A fuzzy logic-based model for the classification of faults in the pump seals of the primary heat transport system of a CANDU 6 reactor

Abstract Many efforts are currently devoted to the development of soft-computing diagnostic tools. Within these efforts, the present work illustrates the application of a fuzzy logic system approach to fault classification in the case of single and multiple failures of different intensity. The approach bears the great advantage that the identification task does not rely on an explicit mathematical model of the behavior of the monitored plant, the if-then rules of the fuzzy classification model being inferred from available preclassified signal data. A case study is presented regarding the classification of faults in the gland seal system of a pump of the primary heat transport system in a pressurized heavy water reactor, Canada deuterium uranium 6.

Predicting the Time To Failure of a randomly degrading component by a hybrid Monte Carlo and possibilistic method

The main goal of a prognostic system is to estimate the Time To Failure (TTF) of a structure, system or component (SSC), i.e. the lifetime remaining between the present and the instance when it can no longer perform its function. Such prediction on the time of loss of functionality is typically done on the basis of measurements of parameters representative of the SSC condition state and degradation process. Uncertainties from two different sources affect the prediction: randomness due to inherent variability in the SSC degradation behavior (aleatory uncertainty) and imprecision due to incomplete knowledge and information on the SSC stress and strength characteristics (epistemic uncertainty). Such uncertainties must be adequately represented and propagated in order for the prognostic results to have operative significance, e.g. in terms of maintenance and renovation decisions. This work illustrates an hybrid Monte Carlo and possibilistic method for the representation and propagation of aleatory and epistemic uncertainties, with reference to a case study of a component which is randomly degrading in time according to a stochastic fatigue crack growth model of literature. The crack depth is assumed as representative parameter of the component condition state and its limit value beyond which failure occurs is assumed to be affected by epistemic uncertainty, as represented by a possibility distribution.

Recognizing signal trends on-line by a fuzzy-logic-based methodology optimized via genetic algorithms

The present work addresses the problem of on-line signal trend identification within a fuzzy logic-based methodology previously proposed in the literature. A modification in the application of the methodology is investigated which entails the use of singletons instead of triangular fuzzy numbers for the characterization of the truth values of the six parameters describing the dynamic trend of the evolving process. Further, calibration of the model is performed by a genetic algorithm procedure. In an example of application of the method, this procedure is also exploited for feature selection, i.e. for choosing which of the measured plant signals are relevant for the transient identification.

Fuzzy-logic-based identification of transients in the feed-water system of a nuclear power plant

This paper addresses the problem of the on-line identification of transients in monitored process signals. A fuzzy-logic-based methodology is adopted. A case study is considered regarding the identification of transients in one section of the feed-water system of a nuclear power plant. The noise associated with the (simulated) signals renders the task of trend identification rather difficult. In spite of this, the method is shown to be capable of capturing important features of the signals trends and of correctly classifying them as increasing, decreasing or steady.