Insulation Permanent Fault Prediction Based On The Instantaneous Fault Signal Analysis In Dis Tribute Network

Insulation degradation of distribution network line is a gradual process.Before insulation permanent fault occurs,it always produce lots of transient fault in each line.It's indeed a very difficult task to predict the circuit's permanent insulation fault happen by utilizing these transient faults.On the one hand,transient fault signals can't be obtained by experiment or simulation,which need a long time period of tracking with troubleshooting records.In the other hand,transient faults' signals can be produced by different kinds of fault reasons,and non-insulated factor transient fault often occurs.In a word,transient fault signals' collecting are disturbed by various factors.If these problems can be solved,then track and analyze those faults,it will be helpful for maintenance personnel to know the development trend of the insulation state of power distribution line equipment and arrange the maintenance plan more reasonably.Line insulation monitoring is based on the fault line selection.This paper firstly analyze the single-phase grounding fault signal in theory and combine the existing fault line selection method based on transient zero mode current with HHT signal processing techniques to extract line selection characteristic quantity.And verify the feasibility of the line selection method by using the simulation.Then,classifies all the transient faults as line's faults by line selection module,and make phased statistical analysis on the fault zero mode current signal in time,frequency and entropy domain.Select appropriate characteristic quantities to describe the fault's severity,and divide those transient faults into three severity areas include mild,moderate and severe failure on the two-dimensional feature plane.Considering that the transient faults occur in distribution line field are affected by all kinds of factors,it is difficult to assess the degree of insulation deterioration.So makes a comprehensive assessment of recent transient faults to assess the condition of line insulation.If severe faults occur in line in several times in succession,it is considered that the feeder aging is serious and the maintenance arrangement is necessary.If slight faults are more than moderate faults and severe faults,it is considered that the feeder aging is mild.Therefore,using the thinking method based on clustering analysis to obtain the Euclidean distance central point of two-dimensional characteristic quantity sample-feature set composed of many transient faults.When the value of insulation deterioration corresponding to the central point over the threshold curve which is set,the operators will be warned immediately,achieving the prediction of insulation permanent fault.The feasibility of this method is verified by field cases.Finally,on the basis of 1-9 scaling method of analytic hierarchy process,this paper raise a new scaling method based on the transient fault's occur-frequency.And utilize this improved analytic hierarchy process to weight assignment for samples.Fault phase also be introduced in the model to make transient faults' degree of membership,which for weighing those fault's degree of being subordinate to insulated or non-insulated.It combines these two methods to achieve weight assignment of the samples comprehensively,and improve the insulation monitoring model.The feasibility of the weight valuation method is verified by comparing the prediction conditions of cases before and after improvement.