The Method Of Medium-Voltage Distruibution Network Fault Diagnosis Based On Data Of Power Distribution And Utilization

Distribution network,as the last link of power transmission,bears the important task of direct power supply to users,and its reliability and power quality greatly affect people's daily life.When a fault occurs in distribution network,rapid fault diagnosis,fault location and isolation of fault components has important significance on shortening the outage time,reducing the areas of power cut,reducing loss of users and improving the reliability of distribution network.At present,most of the distribution network fault diagnosis is for short-circuit fault,such as fault phase,three-phase short-circuit fault.But the research on the disconnection fault especially disconnection fault without grounding is less and lack the effective diagnostic methods.With the deep construction of power distribution and uitilization,many parts of the supply side and demand side information acquisition system have accessed to the integrated data platform,which accumulates a large amount of operation data from power distribution to the use of electricity and provides sufficient data for fault data.When faults happen,there may be relevant,unknown and potentially valuable information in these data.If the useful information can be screened out from the massive data rapidly and potential link can be digged out the between them to realize the accurate judgment of the fault,it can provide the basis for the power grid operation personnel to reduce failure loss and improve the power supply reliability.In this paper,taking the power distribution and utilization of an area in East China as an example,the data of the fault diagnosis of MVdistribution network is divided into three types of data: electrical data,time space data and other data.For the electrical data,according to commonly used fault characteristics determine the useful steady state characteristics of electrical quantities,and for the time and space data,the time and space attributes are calculated from the fault occurrencing time,the fault occurrencing area to extract fault time-space attributes.On this basis,for the disconnection fault without grounding of branch line,an association rule mining method based on data feature selection is proposed.First the continuous features is converted to Boolean feature by chi-square splitting algorithm,at the same time the MSApriori algorithm is used to solve the problem of rare items in the fault information,and then on the basis of this,the kulc rule is applied to eliminate the redundant rules to form the simple rule family,which can be taken as failure criterion.At last the practical numerical analysis based on the historical data of power distribution and utilization information system in an area in East China shows the accuracy of the diagnostic method.For break line fault of distribution network with distributed generators,analyse the change of the fault current especially zero sequence current in MV distribution network under the influence of distributed generators by the use of compound sequence network and symmetrical component method,and discuss the influence of the change on the relay protection device which may have wrong action or reject action.Therefore,consider the comprehensive use of electrical data and switching data for fault diagnosis.Firstly apply Petri net to switch information to generate switch fault degree,and use the electrical characteristics to generate electrical fault degree.On the basis of the above methods,the improved D-S evidence theory model is used to fuse the switch fault degree and electrical fault degree to obtain the final diagnosis results.At last,the feasibility of the method is proved by a practical example.