DESIGNING AN IMPROVED NEURAL NETWORK FOR THE EARLY DETECTION OF ANOMALIES IN NUCLEAR POWER PLANTS

The effectiveness and dependability of these vital energy infrastructures depend heavily on the early detection of anomalies in nuclear power plants (NPPs). Anomalies in a plant's operations might be signs of the impending equipment failure, a danger to workers' safety, or departure from ideal performance, all of which call for quick attention and preventative actions. Traditional NPP monitoring methods depend on the human inspections and predetermined thresholds, which are only sometimes successful in picking up the complicated irregularities. This Study introduces a new, Improved Bat and Grey Wolf Optimized Recurrent Neural Network (IBGWO-RNN) approach to detect the anomalies in NPPs. In this case, the RNN classification effectiveness is increased by using the IBGWO method. The American Nuclear Society ANSI / ANS-3.5 Nuclear Simulator Standard dataset has been used to assess the success of the suggested approach. Each input feature vector will be normalized by using the Z-score Normalization. A Kernel Principal Component Analysis (KPCA) is performed to extract the properties from segmented data. The results of the research show that the recommended methodology beats earlier approaches in terms of the Accuracy, Precision, Recall, and F1-score. Our suggested approach advances anomaly identification, resulting in safer and more effective operations for NPPs.