Online Compensation Method for Temperature Measurement Error of Core Outlet Thermocouple

Thermocouple readings are critical for monitoring core anomalies and play a vital role in preventing damage to core components and major safety incidents. However, thermocouples at the core outlet often have an installation position deviation, resulting in significant temperature measurement errors. This paper first analyzes the factors influencing thermocouple temperature measurement errors, then proposes a computational model based on Random Forest (RF), and finally employs the Crested Porcupine Optimizer (CPO) to optimize the RF hyperparameters. The research results demonstrate that the proposed method outperforms existing approaches in predicting temperature errors, achieving a maximum post-compensation thermocouple measurement error of 0.52% while maintaining per-sample computation time at the microsecond level. This method is highly significant for preventing and avoiding nuclear power plant safety accidents, as well as advancing the application of machine learning methods in nuclear engineering.