Simultaneous Fault Diagnosis of the Reactor Coolant System Based on the DSM Evidence Theory

Abstract

Fault diagnosis of nuclear power plant (NPP) has become a major research topic in ensuring the reliability and safety for the NPP operation. In the light of complexity and information uncertainty of nuclear reactor coolant system, some diagnosis techniques based on the multi-sensor information fusion theories have been widely used. As one part of the those theories, the Dempster-Shafer (D-S) theory excelled in demonstrating and combining uncertain information has witnessed a wide range of applications in the fault diagnosis fields. It can be inferred from the previous studies that the fault diagnosis method based on the D-S theory is efficient; However, this method is mainly intended for single fault diagnosis. In fact, in practical application, simultaneous faults often occur. Based on the DSm theory (Dezert-Smarandache theory of plausible and paradoxical reasoning), this paper proposes a simultaneous fault diagnosis method, which is intended to deal with the simultaneous faults in the reactor coolant system and adequately cope with the sensor information uncertainty. This method can be divided into four steps. Firstly, with the features of the simultaneous fault considered, a diagnosis model of the simultaneous fault is designed, which is based on the DSm theory. Secondly, for the instant information of each senor, the fuzzy membership degree is used to set the corresponding basic probability assignment (BPA) function for each sensor. Thirdly, the above-mentioned BPA functions are combined by utilizing the classic DSm rules. Finally, a decision on the basis of maximum support rule and absolute support rule is made to determine the real fault modes. After that, the experiments on the test-bed are conducted to prove the efficiency of this method.