Study On The On-Line Monitoring And Fault Diagnosis Technology Of Dry Air Core Reactor

With the characteristics of the good linearity,loss efficient,low noise and easy maintenance,dry air core reactors are widely used in power system with the proportion of 70% in the transmission and distribution grid of 66 kV and below voltage.With more and more dry type air core reactors put into operation,faults happened continuously.The relevant research show that the inter-turn short circuit accounts for more than 60% of dry air core reactors faults,which can't be monitored immediately by the existing measurement methods.The low sensitivity and poor reliability may delay the warning then lead to fire and blast of the reactors.Therefore,it is of great significance to find a on-line monitoring and fault diagnosis technology of high sensitivity for inter-turn short circuit faults of reactors.The equivalent circuit of dry air core reactors under normal operation and inter-turn short circuit faults conditions are set up,through which we can get the equivalent resistance,reactance and power factor.Then,the monitoring method for identifying low-turn short circuit faults of dry air core reactors is proposed,based on the change of equivalent resistance,reactance and power factor of reactors under normal operation and inter-turn short circuit faults conditions.According to the structure of power grid,this paper designs an on-line monitoring system for dry air core reactors.The main processor used the chip combination of ARM and FPGA.FPGA realized the synchronous acquisition of three-phase voltage and current of reactor and frequency measurement.On the one hand,ARM is used to processing and computing data,then calculate the electrical parameters through embedding algorithm.On the other hand,it upload the data to expert software by Ethernet.The expert software has the functions of centralized monitoring,management and information query.What's more,the state information of each dry air core reactor is obtained by embedding fault diagnosis algorithm,and the related electrical parameters are displayed by chart.The accuracy measurement test,inter-turn short circuit fault simulation test of reactor and electromagnetic compatibility test are carried out by the test platform.The equivalent resistance,reactance and power factor of three-phase reactor under fault conditions can be obtained.The test result show that measuring accuracy and reliability of the device can meet the requirements.The generalized regression neural network(GRNN)and BP neural network model are established via the related parameters measured by the monitoring device.The diagnostic accuracy of GRNN model is 94%,while the BP neural network model is 86%.The diagnostic results show that the diagnostic accuracy of GRNN model is significantly higher than that of BP neural network.Therefore,GRNN model is applied to expert software in this paper.Finally,the system has been successfully installed and operated in the substation.The field operation results show that the on-line monitoring system for dry air core reactor designed in this paper is feasible.There are 37 figures,11 tables and 57 references in this paper.