Predictive Diagnostics of Control Sensors in a Nuclear Reactor

Abstract

In this paper, we propose an approach for the pinpoint detection of inoperable sensors based on the analysis of diagnostic parameters. The “sliding correlation” coefficients, the fourth eigenvalue, and the relative deviation of the reconstructed sensor reading may be such parameters. With the help of the above diagnostic parameters, it is possible to immediately determine the fuel assembly in which one of the direct charge sensors is presumably out of order, then analyze the level correlations between each other, and, using the restored value, make a final judgment on the performance of each altitude sensor. The developed software makes it possible to carry out an express analysis of the archive of the VVER reactor. This has scientific and practical value in the context of improving the quality of work of operational personnel and analyzing situations that require additional attention and more detailed analysis. The application of this approach can help detect malfunctions in sensors and take timely measures to prevent possible problems and emergencies, which is an important step in ensuring the safety and efficiency of the VVER reactor monitoring and control system.