Condition Monitoring And Defect Diagnosis System Of Power Capacitor In Substation

Based on neural network and information fusion,an on-line monitoring system for power capacitor is designed in this paper.Firstly,the research background,necessity,research status and development trend of on-line diagnosis of power capacitor are introduced.It is necessary to monitor the operation state of power capacitor,find out its defects and faults in time and deal with them as early as possible,so as to avoid the possible serious consequences caused by fault accumulation.The existing diagnosis method is preventive test,that is,the capacitor should be maintained regularly.Although this method can achieve the purpose of maintenance to a certain extent,it also has some defects,so the trend of maintenance is from regular maintenance to condition based maintenance.Secondly,the neural network and information fusion are applied to the fault diagnosis of capacitor.Through the use of transformer and modeling,the normal operating parameters of capacitor,including voltage,current,reactive power,ambient temperature and humidity,capacitor temperature,are extracted for information fusion diagnosis.Then the overall structure and communication subsystem of the condition monitoring and diagnosis system are designed,and the hardware and software are configured.The system includes main station management layer,local monitoring layer and station end management layer.The main station management layer meets the user interaction through the server and workstation,and also realizes the monitoring and data collection of the running state of the capacitor.It monitors the operation of the local detection layer and analyzes the online monitoring data of the capacitor by the mobile phone,and gives an alarm in time in case of any fault.The station management layer consists of an industrial control panel and a 1Gbps Ethernet switch to realize the measurement control,communication management,data collection,processing analysis and display of all local monitoring units in the substation.When the index exceeds the limit,the capacitor that is about to fail is diagnosed online and early warning is given,and the results and equipment status are transmitted to the main station system or remote system through the network.The main equipment of the local monitoring layer includes electrical measurement unit and temperature measurement unit,which areresponsible for real-time monitoring of indoor temperature,humidity,voltage size,voltage quality and voltage waveform distortion rate.Under normal circumstances,the local monitoring layer uploads data every minute.Finally,the system is tested and analyzed in practice.The practice shows that the relevant monitoring parameters of the established condition monitoring system meet the accuracy requirements of defect diagnosis for electric and non-electric information;the designed alarm function operates normally;the designed non parameter diagnosis method can track the heating behavior of capacitor;the designed parameter based diagnosis method provides more decision-making basis for information fusion diagnosis.As there is no capacitor fault in the test run,the information fusion method described in Section 2.4 has not been well tested and verified.