Nuclear Power Plant Control System Health Index Based on Combination Weighting of CPA Indexes

Safety of nuclear power plants is of vital importance. In this context, Control Performance Assessment (CPA) emerges as a valuable tool for the evaluation of control system performance within nuclear power plants, employing a range of pertinent performance indexes. A comprehensive health index for control systems can be obtained by applying weighted considerations to these indexes. This paper introduces a novel combination weighting approach for the CPA indexes applicable to control systems in nuclear power plants. In situations that historical abnormal data for control systems is insufficient, entropy serves as a weighting mechanism. Moreover, for systems characterized by prolonged degradation processes, a hybrid weighting strategy is proposed. This hybrid approach combines entropy weighting with weights determined through SHAP (SHapley Additive exPlanations), which quantifies the significance of various indexes in anomaly diagnosis. The proposed weighting method has shown good performance through testing with real operational data of nuclear power plants.