Prediction method for thermal-hydraulic parameters of nuclear reactor system based on deep learning algorithm. (September 2021)

• Record Type: Journal Article
• Authors: Lu, Qi; Yuan, Yuan; Li, Fengchen; Yang, Bo; Li, Zhe; Ma, Yu; Gu, Yiyu; Liu, Dingming
• Is Part Of: Applied thermal engineering. Volume 196(2021)
• Publication Date: 2021-09
• Subjects: Nuclear reactor core -- Once-through steam generator -- Thermal-hydraulic design -- Deep learning algorithm -- Neural network; Heat engineering -- Periodicals; Heating -- Equipment and supplies -- Periodicals; Periodicals; 621.40205
• DOI: 10.1016/j.applthermaleng.2021.117272
• Languages: English
• ISSNs: 1359-4311

Abstract

Highlights: A prediction program for thermal-hydraulic parameters of nuclear reactor system was established by deep learning algorithm. Deep learning algorithm was mainly composed of neural network, activation function, error function, optimization algorithm and data initialization modules. KLT-40S nuclear reactor core and sleeve once-through steam generator were selected as prediction objects with good agreement results. Abstract: The nuclear reactor core and the steam generator are the core components of nuclear reactor system, which directly determine the heat releasing and carrying capacity of nuclear reactor system. The traditional thermal-hydraulic design of nuclear reactor system generally selects the linear iterative mode, which requires the repeated "calculation-evaluation-correction". The whole process is tedious and time-consuming, which also relies on the design experience. In recent years, the successful application of artificial intelligence technology in many fields has brought new enlightenment to the innovation for the thermal-hydraulic design of nuclear reactor system. In this paper, a prediction program for the thermal-hydraulic parameters of nuclear reactor system was established by means of deep learning algorithm, which was mainly composed of the neural network module, the activation function module, the error function module, the optimization algorithm module and the data initialization module. Then, the KLT-40S nuclear reactor core and the tube-in-tube once-through steam generator were selected as the analysis objects, and the corresponding input card of RELAP5/SCDAPSIM program was built. Based on the prediction program, the calculation results of RELAP5/SCDAPSIM program were studied, which could realize the rapid prediction for the thermal-hydraulic parameters of nuclear reactor system, and were in good agreement with the calculation results of RELAP5/SCDAPSIM program.