Development of three methods for control rod position monitoring based on fixed in-core neutron detectors

Nuclear reactor core power distribution on-line monitoring system is very important in core surveillance, and this system should have the ability to indicate some abnormal conditions, such as the unacceptable control rod misalignment. In this study, the methodologies of radial basis function neural network (RBFNN), group method of data handling (GMDH) and Levenberg-Marquardt (LM) algorithm are utilized separately to unfold the control rod position from the fixed in-core neutron detector measurements. For using these methods, a large number of in-core detector signals corresponding to various known rod positions are needed. These data can be generated by an advanced core calculation code. In this study, the neutronics code SMART was used. The simulation results show that all these methods can unfold the control rod position accurately, and the performance comparison shows that the regularized RBFNN performs best. Two correction strategies are proposed to correct the simulated fixed in-core detector signals and improve the rod position monitoring accuracy when there are mismatches between actual physical factors and modeled physical factors.