Experimental Study and Intelligent Prediction on Pressure Fluctuation of Accumulator Under Ocean Conditions

Abstract

Liquid sloshing will occur in liquid storage tanks such as accumulator of floating nuclear power plant (FNPP) subjected to additional inertial forces under motion conditions. The study carried out measurement experiments based on the 6-DOF platform to study the sloshing characteristics and pressure variation rule of the accumulator. The results show that surging will induce many kinds of nonlinear free surface sloshing forms, it can be seen that the law of pressure variation is mainly dominated by natural frequency and excitation frequency based on time and frequency domain analysis. Then the study combines the automatic encoder and extreme learning machine to build the deep extreme learning machine (DELM) network to predict the pressure in time series. Based on the phase space reconstruction of the time sequence, the pressure results of the next time are output after the last 15 pressure data are input. The prediction results show that the DELM model has fast speed and high precision and the predicted value is in good agreement with the experimental data. So this study can provide a reference for the pressure monitoring and the artificial intelligence application of FNPP.