Fault diagnosis of cooling pipeline system

Abstract

It is difficult to identify the monitoring parameters of the cooling pipeline system with seawater as medium when the pipeline is partially blocked and small leakage occurs. The control of condenser vacuum leads to the action of the control valve, which will make the system produce output response. Through theoretical analysis, the output response of the pipeline in different working conditions is different, which makes fault diagnosis possible. Therefore, this paper uses Flowmaster software to establish the models of normal working conditions, partial blocking working conditions and small leakage working conditions. The pressure variation in front of the condenser is simulated, and it is found that the pressure variation under the three working conditions is quite different, which indicates that the fault diagnosis can be carried out by identifying the pressure waveform. 400 groups of experimental data are obtained through simulation. Then the fault diagnosis system established by two different input methods of BP neural network is compared, and it is found that the iterations of taking the time series as the network input are less, the accuracy is as high as 96%.