Genetic algorithm-based wrapper approach for grouping condition monitoring signals of nuclear power plant components

Abstract

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.

Fault Detection in Nuclear Power Plants Components by a Combination of Statistical Methods

“…Based on previously published works [29], a condition monitoring scheme capable of identifying the signal groups by a hybrid approach based on CA and GA has been adopted. This approach has been selected because of its capability to limit fault propagation with respect to a CA-based filter approach, and to demand a lower computational effort with respect to a GA-based wrapper approach.…”

“…The latter uses a search algorithm, e.g., Genetic Algorithms (GAs) [29], Differential Evolution [30], etc., as a wrapper around the signal reconstruction model [31]. In this paper, we use a hybrid approach, i.e., filter and wrapper, based on non-overlapping grouping given by Correlation Analysis (CA) and GA [29] (Fig. 2, bottom left).…”

“…2, bottom left). The motivation of the choice is that the GA-based wrapper approach allows finding better performing groups for the specific reconstruction model used, while the contribution of CA leverages the computational burden of the GA (for further details, the interested reader should refer to [29]). To balance false and missed alarms in the fault detection, it is possible to perform an analysis of the residuals to declare whether a component fault has occurred (Fig.…”

“…The 46-dimensional patterns (5798) have been divided into a set X M of 2798 patterns used to perform the hybrid signal grouping (Section 3), and a validation set X V of 3000 patterns used to validate the condition monitoring and fault detection scheme (Section 4). In [29], it has been shown that this hybrid method leads to reconstructions that are more tolerant to the fault propagation problem (that has been mentioned in the Introduction) than proceeding with the reconstruction based on the single group of all signals [19], [27], or on groups of signals defined by a filter CA approach only [19], [25], or by a wrapper GA-based approach only [28].…”

“…The hybrid approach consists of using CA to identify groups of highly correlated signals, and GA to define the group assignment of the remaining signals [29]. Application to the case study of interest leads to CA identifying 5 groups, with one formed by 24 highly correlated signals, and GA searching chromosomes of 22 elements, each one corresponding to one of the 22 signals not assigned to a group by CA.…”