Nuclear accident identification for a PWR nuclear power plant using the Cuckoo optimization algorithm with do not know response generation

Nuclear Power Plants (NPPS) are facilities characterized by complex systems and rigorous inspections. Controlled by human operators in the control rooms, monitoring and managing these plants, assuming responsibility for adhering to protocols that ensure safety, accident prevention and mitigation. In the event of an abnormal occurrence, the control room team must accurately identify the situation and promptly follow established protocols. Precision and timeliness in executing these procedures are essential to ensure the optimal functioning of NPPs. Given the potential serious consequences of misidentification during abnormal events in a Nuclear Power Plant (NPP), different methodologies have been proposed to address the Nuclear Accident Identification Problem (NAIP). This work presents a new methodology for NAIP incorporating a 'Don't Know' response, utilizing the Cuckoo Optimization Algorithm (COA). Which will be referred to NAIP-COA method. The COA is employed to identify the most suitable representative vector, composed of prototype vectors representing various plant states in the model. The prototype vector serves as a representation of a pseudo Voronoi Vector representing a region within a search space. The incorporation of the 'Don't Know' response is grounded in the definition of the influences area of the prototype vector, found by the algorithm, and identifying if an anomalous event is within this region or not. To study the performance of the NAIP-COA method, a case involving 8 simulated operational scenarios (comprising 7 postulated accidents and normal operation) for a Pressure Water Reactor (PWR) Nuclear Power Plant was used in the experiments. The results of the NAIP-COA method were compared with the results in the literature, achieving 99,37% accuracy in the total. (author)