An intelligent active fault tolerant control strategy for pressurized water reactor: Addressing sensor faults

Abstract

Robust and reliable control strategies are essential for ensuring the safe and efficient operation of thermal energy systems, particularly in nuclear reactors. This paper proposes an intelligent active fault-tolerant control strategy to address sensor faults in pressurized water reactors. The proposed framework comprises three key components: a control mechanism, a fault detection and diagnosis unit, and a reconfiguration mechanism. A Proportional-Integral-Derivative controller, optimized using ant colony optimization, is implemented to regulate neutron density. The fault detection and diagnosis unit employ a recurrent neural network estimator to detect sensor faults by analyzing residuals. Upon fault detection, a reconfiguration module, which is also based on a recurrent neural network, predicts the true value of the faulty sensor using correlated measurements from other sensors. The prediction accuracy is quantified using the mean square error performance index. The effectiveness and practical applicability of the proposed strategy are demonstrated through simulation studies, including a representative drift fault scenario.