Research on the soft measurement of the flow by reactor coolant pumps based on the motor current signature analysis

Abstract

Effective monitoring and identification of the operational flow of reactor coolant pumps (RCPs) are crucial for enhancing the safety and stability of nuclear power operations. To achieve the goal without the potential interference from intrusive sensors, the paper employs wavelet analysis and intrinsic time scale decomposition (ITD) to examine the collected current signals in both the time and frequency domains. The recurrent neural network (RNN) is used to determine the operational status of the pumps at different flow rates. The soft measurement model of the RCP flow, based on motor current signature analysis (MCSA), is built and experimentally validated using the RNN to identify the pump’s performance at different flow rates. The results show that the dominant frequency of pressure pulsation corresponds to the axial frequency in impeller channels, with the amplitude of the frequency component exhibiting a direct positive correlation to the flow rate. The impeller channel-generated pressure pulsation signal’s dominant frequency corresponds to the blade passing frequency, which exhibits an amplitude increase with higher flow rates. The soft measurement model, after the RNN training, achieves a recognition accuracy as high as 95.3% for flow rates of 0.8Q_d, 0.9Q_d, 1.0Q_d, 1.1Q_d, and 1.2Q_d. The model provides a certain value for subsequent studies in the field of pump flow monitoring.