Prediction of Sensor Data Accuracy in Thermal Experimental Benches Using GRU-GCN Neural Network Model

作者：Linjun Yang, Tong Li, Yongchao Liu, Bo Wang, Jiangkuan Li, Jiming Wen, Sichao Tan, Ruifeng Tian · 发表于：Volume 6: Thermal-Hydraulics and Safety Analysis · 年份：2024 · DOI：10.1115/icone31-134843

As a large-scale facility with multi-system coupling, the operation safety of nuclear power plant is of great importance to the environment and human society. Accurate monitoring of the overall operation status of nuclear power plants (NPP) is not only the key to ensure its safe operation, but also an important means to improve energy efficiency and reduce potential risks. In order to solve the problem of simultaneous prediction of multiple operating parameters of nuclear power plant, this study proposes a deep learning model combining Graph Convolution Network (GCN) and Gated Cycle Unit (GRU), which have the advantages of processing complex network data and time series data respectively. In order to verify the validity of the proposed model, a series of experiments are carried out on the thermal test-bed simulating the operation characteristics of nuclear power plants. The experimental results show that compared with the existing methods, GRU-GCN model has significantly improved the prediction accuracy. This result not only shows the potential of the deep learning model combining GCN and GRU in processing complex system data, but also provides a new idea for the safety monitoring and operation management of nuclear power plants in the future.