A novel ensemble Model Aggregation for Robust Signal Reconstruction in Nuclear Power Plants Monitoring Systems

Validity and accuracy of sensor signals are crucial for the enhancement of the safety, reliability and performance of complex industrial systems such as nuclear power plants. In this view, on-line sensor monitoring aims at detecting anomalies in sensors and reconstructing their correct signals during operation. Auto-associative regression models are commonly adopted to perform the signal reconstruction task. Nevertheless, on real scale applications the number of sensors signals is too large to be handled effectively by one single model. To overcome this problem, one may resort to an ensemble of reconstruction models, each one handling an individual (small) group of sensor signals. The outcomes of the models need then to be opportunely combined. In this work, three methods for aggregating the outcomes of the individual reconstruction models of a randomized ensemble are implemented, applied and compared on a case study concerning the reconstruction of 920 simulated signals of the Swedish Forsmark-3 Boiling Water Reactor (BWR). A novel procedure is developed to combine two of these methods in such a way to mostly benefit from their capabilities.