On-line sensor monitoring aims at detecting anomalies in sensors and reconstructing their correct signals during operation. The techniques used for signal reconstruction are commonly based on auto-associative regression models. In full scale implementations however, the number of sensors to be monitored is often too large to be handled effectively by a single reconstruction model. In this paper we propose to tackle the problem by resorting to a pool (ensemble) of reconstruction models, each one handling an individual group of signals. This approach involves two main technical steps: firstly, a procedure for constructing signal groups, and secondly a procedure for combining the outputs of the reconstruction models associated to the groups. For the signal grouping step, a wrapper optimization search is proposed to identify the optimal number of groups in the ensemble and the size of the groups. For the model output aggregation step, a simple arithmetic average is adopted. Ensemble accuracy and robustness is achieved by promoting diversity between the signal groups through the use of the Random Feature Selection Ensemble (RFSE) technique in combination with the Bootstrapping AGGregatING (BAGGING) technique for training data selection. The individual reconstruction models are based on Principal Components Analysis (PCA). The proposed approach has been applied to a real case study concerning 215 signals monitored at a Finnish nuclear pressurized water reactor. The results obtained have been compared with those achieved by an equivalent ensemble of models based on a grouping directly optimized by a Multi-Objective Genetic Algorithm (MOGA).

Detecting anomalies in sensors and reconstructing the correct values of the measured signals is of paramount importance for the safe and reliable operation of nuclear power plants. Auto-associative regression models can be used for the signal reconstruction task but in real applications the number of sensors signals may be too large to be handled effectively by one single model. In these cases, one may resort to an ensemble of reconstruction models, each one handling a small group of sensor signals; the outcomes of the individual models are then combined to produce the final reconstruction. In this work, three methods for aggregating the outcomes of a feature-randomized ensemble of Principal Components Analysis (PCA)-based regression models are analyzed and applied to two case studies concerning the reconstruction of 215 signals monitored at a Finnish nuclear Pressurized Water Reactor (PWR) and 920 simulated signals of the Swedish Forsmark-3 Boiling Water Reactor (BWR). Based on the insights gained, two novel aggregation procedures are developed for optimal signal reconstruction.

On-line sensor monitoring allows detecting anomalies in sensor operation and reconstructing the correct signals of failed sensors by exploiting the information coming from other measured signals. In field applications, the number of signals to be monitored is often too large to be handled effectively by a single reconstruction model. A more viable approach is that of decomposing the problem by constructing a number of reconstruction models, each one handling an individual group of signals. To apply this approach, two problems must be solved: (1) the optimal definition of the groups of signals and (2) the appropriate combination of the outcomes of the individual models. With respect to the first problem, in this work, Multi-Objective Genetic Algorithms (MOGAs) are devised for finding the optimal groups of signals used for building reconstruction models based on Principal Component Analysis (PCA). With respect to the second problem, a weighted scheme is adopted to combine appropriately the signal predictions of the individual models. The proposed approach is applied to a real case study concerning the reconstruction of 84 signals collected from a Swedish nuclear boiling water reactor. (C) 2010 Elsevier B.V. All rights reserved.

Monitoring of sensor operation is important for detecting anomalies and reconstructing the correct values of the signals measured. This can be done, for example, with the aid of auto-associative regression models. However, in practical applications, difficulties arise because of the need for handling large numbers of signals. To overcome these difficulties, ensembles of reconstruction models can be used. Each model in the ensemble handles a small group of signals and the outcomes of all models are eventually combined to provide the final outcome. In this work, three different methods for aggregating the model outcomes are investigated and a novel procedure is proposed for obtaining robust ensemble-aggregated outputs. Two applications are considered concerning the reconstruction of 920 simulated signals of the Swedish Forsmark-3 Boiling Water Reactor (BWR) and 215 signals measured at the Finnish Pressurised Water Reactor (PWR) situated in Loviisa.

In nuclear power plants it is important that sensor failures be detected and accommodated before significant performance degradation results. In this work, the task of reconstructing the correct signal from a faulty sensor is addressed by resorting to an ensemble of reconstruction models, each one handling an individual (small) group of sensor signals. Operatively, diverse signal groups are randomly generated according to the Random Feature Selection Ensemble (RFSE) technique and a corresponding number of Principal Components Analysis (PCA)-based regression models are built using the groups' signals. The outcomes of these models are combined by three different techniques whose performances are compared on a real case study concerning 215 signals monitored at a Finnish nuclear pressurized water reactor.

This paper investigates the still largely unexplored issues of redundancy and diversity as seen in the context of condition monitoring systems, and more specifically in on-line signal validation systems. We believe these aspects should play a more central role in the development of such systems so that the potential for common cause failures of the models used for condition monitoring is minimised. Accurate monitoring of operating conditions can have a significant impact on the operation of modern plants, with respect to production, accident management and maintenance. Such monitoring is based on a collection of sensors. During plant operation, some sensors may experience anomalies (e.g. drifts or failures), which might convey inaccurate or misleading information about the actual plant state to automated controls and to the operators. A robust monitoring system must be able to detect such anomalies and reconstruct correctly the signals of the failed sensors, as done for example in the PEANO system developed at the OECD Halden Reactor Project [1]. On-line sensor monitoring evaluates instrument channel performance by assessing its consistency with other plant indications (signal validation). Information about the condition, performance and calibration state of the channels through accurate and frequent monitoring while the process is in operation provides a basis for determining when signal reconstructions and sensor recalibrations are necessary. In real systems, the number of sensor signals involved in the monitoring of the plant state is generally too large to be handled effectively within a single validation and reconstruction model. One approach to address the problem is to create smaller groups of signals and develop one model for each group [1, 2], thereby significantly reducing the modelling complexity. In this work, signal grouping is carried out by means of Genetic Algorithms (GAs) [3-8]. GAs are global search algorithms that try to find optimal solutions, as defined by a set of explicit objective functions. A signal grouping algorithm for generating an effective set of signal groups should be based on a set of objective functions for grouping optimization that express a variety of desirable properties. In particular, objective functions can be defined to capture the following aspects: • Mutual information content of the signals in a group, so that accurate signal validation models can be built. • Accuracy of the models adopted to validate and reconstruct the signals in the groups. • Size of the groups, so that the signal validation modelling process is both computationally feasible and efficient. • Redundancy and diversity of the groups, i.e. having signals represented in a number of groups (redundancy) and in groups with a varied signal composition (diversity). • Completeness, so that a large majority of the signals considered is validated. Of particular interest for the topic of this paper is the introduction of objective functions that favour the construction of redundant and diverse signal groups. By having redundant signal validation, i.e. having each signal validated and reconstructed by a relatively large number of models derived from overlapping signal groups, one can significantly reduce the effect of modelling variance and achieve a more robust signal validation and reconstruction performance. Diverse signal validation can also be achieved through signal grouping by enforcing that the composition of redundant signal groups is as varied as possible, thereby reducing the possibility of common cause validation failures which could arise if a signal was to be validated and reconstructed by a uniform selection of related signals. The requirement for diversity can also be implemented by enforcing diversity in the modelling techniques used to build the signal validation models, thereby reducing the possibility for common cause failures that could derive from the adoption of a single signal validation technique.

On-line sensor monitoring allows detecting anomalies in sensor operation and reconstructing the correct signals of failed sensors, Since in field applications the number of signals to be monitored is often too large to be handled effectively by a single reconstruction model, a more viable approach is that of decomposing the problem by developing a number of reconstruction models, each one handling an individual group of signals. In this paper, Multi-Objective Genetic Algorithms (MOGAs) are devised for finding the optimal groups of signals used for building reconstruction models based on Principal Component Analysis (PCA). A weighted scheme is adopted to combine appropriately the signal predictions of the individual models. The proposed approach is applied to a real case study.

Sensor validation is aimed at detecting anomalies in sensor operation and reconstructing the correct signals of failed sensors, e.g. by exploiting the information coming from other measured signals. In field applications, the number of signals to be monitored can often become too large to be handled by a single validation and reconstruction model. To overcome this problem, the signals can be subdivided into groups according to specific requirements and a number of validation and reconstruction models can be developed to handle the individual groups. In this paper, multi-objective genetic algorithms (MOGAs) are devised for finding groups of signals bearing the required characteristics for constructing signal validation and reconstruction models based on principal component analysis (PCA). Two approaches are considered for the MOGA search of the signal groups: the filter and wrapper approaches. The former assesses the merits of the groups only from the characteristics of their signals, whereas the latter looks for those groups optimal for building the models actually used to validate and reconstruct the signals. The two approaches are compared with respect to a real case study concerning the validation of 84 signals collected from a Swedish boiling water nuclear power plant.

A robust monitoring system must be able to detect also the faults possibly occurring in the sensors and to reconstruct correctly the signals monitored by the sensors detected as failed. To this aim, it is important to identify the appropriate groups of signals which carry sufficient information for reconstructing the signal(s) of the failed sensor(s) with the required accuracy. In this paper, the problem of finding an optimal grouping of signals for sensor validation is handled by means of a Multi-Objective Genetic Algorithm (MOGA) optimization. Two objectives are considered: the maximization of the correlation between signals and the maximization of the group size. The approach is applied to a data set of 84 signals collected from a boiling water nuclear power plant located in Oskarshamn, Sweden.

On-line calibration monitoring evaluates the performance of instrument channels by assessing their mutual consistency and possibly their consistency with other plant measurements. Experience at several nuclear power plants has shown this overall approach to be very effective in identifying faulty instrument channels. Most applications to date have however been confined to the monitoring of a relatively small number of instrument channels. Even though these applications have demonstrated the calibration monitoring properties of the applied techniques, questions remain open on the scalability of the same techniques to large-scale applications, where by large-scale we mean plant-wide implementations involving several hundreds if not thousands of instrument channels. In this paper, we propose a number of prospective solutions grouped in two main categories: i) solutions to handle the calibration monitoring of a very large number of instrument channels and ii) solutions to handle the ensemble of models that might derive from the decomposition of the calibration monitoring task.