Study On Characteristics And Recognition Methods Of Blockage And Leakage Of Cooling Water System

Fault identification and processing of large water cooling system is important for the system long-term health and its high efficiency operation.We can achieve a large number of real-time processing parameters by the application of sensor technology and modern communication technology.In this paper,the characteristics of Pipe network blockage and leakage and their identification methods are studied in order to make full use of the data acquired by the data acquisition system,to discover system failures more quickly and to make more reasonable diagnosis.In this paper,the structural characteristics,resistance distribution,operation regulation and failure of large cooling system are analyzed.Proposed a large-scale data acquisition solution,which allows physical channel changes.Achieve more than 400 signals per second data acquisition by using lower cost data acquisition equipments.According to the running characteristics of the cooling system,the method of using the threshold of the cooling system is established,and the cooling system monitoring software is designed.Reliability of the monitoring software is validated in practical engineering applications.Pipe clogging caused by the problem of the filter failure is studied using the signal processing method.Wavelet transform is used to remove the data noise and get time series data of the flow and impedance coefficient more accurate.By comparing the flow,resistance and impedance of the branch of the filter in the normal working and fault conditions,a method of identifying the failure of the filter based on the flow or pipeline impedance time series is proposed.In order to diagnose the fault location of the whole network,the variation law of the hydraulic parameters of the pipe network is studied by simulating the fault condition.The simulation results show that the pressure offset and its direction before and after the clogging point have distinct characteristics,which can be used as the basis for identifying the clogging position.For the problem of pipe network leakage,this paper puts forward the equivalent pipe network model for the leakage condition,and provides a theoretical explanation for the analysis of the leakage law.By the method of simulation,the leakage law of the pipe network in the form of sub-catchment is analyzed,and the conclusions are drawn that Leakage at different leak point under the same leak rate has different effects on the operating parameters of the water system.In order to determine the location of the water leakage point under the leakage condition,a two-stage BP neural network leakage location model is established by taking the pressure offset of the pipe network as the characteristic.Multi-level division of the complex pipe network can control the network input Dimension,to build a network with strong generalization ability and easy training.The model is simulated by multi-level branch pipe network.The results show that the neural network fault diagnosis model can reliably detect the location of the leak.