International Symposium on Symbiotic Nuclear Power (ISSNP2024)

With the development of nuclear energy, small modular reactor (SMR) has outstanding performance in safety characteristics and multi-purpose.The safety of nuclear power plant has always been concerned. Fault diagnosis can help nuclear power plant accurately predict and identify faults and ensure the safe and stable operation of the system. In this paper, the fault diagnosis model of small modular pressurized water reactor (PWR) was established by using several intelligent algorithms based on the simulation data, and the diagnostic effect of each model was analyzed. Firstly, the normal and accident samples of the system operation were obtained based on the simulation model of the small modular PWR. The sensitive features of fault diagnosis model of small modular PWR were extracted. Then, fault diagnosis model was established based on artificial intelligence algorithm, including BP neural network, support vector machine (SVM), convolutional neural network (CNN), random forest four typical artificial intelligence classification algorithms. The fault diagnosis model was used to identify the fault categories of the system, and the diagnosis effect of each model was analyzed respectively. The results show that the diagnostic effect of random forest model was the best, followed by BP neural network model and convolutional neural network model, while the diagnostic effect of support vector machine model was relatively average.
 

Small Modular PWR, Fault Diagnosis, Artificial Intelligence Algorithms