Research On Overhead External Fault Indicator Based On Data Mining

With the development of power system,China’s distribution network is characterized by complicated lines,changeable environment,frequent failures and high maintenance cost.In the distribution network,the voltage change will affect the service life of all kinds of electrical equipment.When the grid voltage fluctuates,whether the voltage changes too much or too little will lead to the damage of related electrical equipment,which will lead to the occurrence of line faults.Therefore,in order to ensure the safe and stable operation of the distribution network,how to accurately locate the fault points in the power grid lines has become an important problem that the power grid maintenance personnel need to solve urgently.In the research of distribution network fault location,fault indicator is widely used because of its flexibility and low cost.At the same time,because of the huge amount of data acquired by the distribution network running state detection equipment,it is difficult to analyze the fault data manually if the information acquisition equipment acquires it.Therefore,it is an important method to effectively use data mining technology in the era of big data to solve massive data mining.Data mining technology can realize the mining of effective information in massive data,the mining of internal information of data and the display of its results.In view of this,this paper studies a new type of overhead fault indicator,which can accurately locate the fault source by distance through data mining technology.Firstly,the fault types of distribution network,parameters of fault indicators,performance of data collector,signal measurement and control chip are studied,and the overall scheme design is carried out according to the fault diagnosis principle of fault indicators.Secondly,the hardware and software of the overhead fault indicator are designed.Firstly,the design scheme and basic functions of the fault indicator are introduced from the overall structure,and the detailed circuit design of each hardware and software module is carried out.Then,a simulation model of distribution network fault detection is established,and different types of fault distances are set in the model.In different types of scenarios,the load and fault resistance are adjusted several times to obtain massive data sets needed for fault detection research of distribution network.Further,the neural network algorithm in data mining technology is used to locate the fault.Finally,the contrast experiment shows that BP neural network can effectively locate the fault distance of distribution network and accurately identify the fault source.It is verified that the fault location method based on data mining technology can accurately locate the fault of distribution network,and realize real-time monitoring and intelligent remote fault detection of distribution network.It has certain application value and popularization value in practical engineering.