| Due to different construction times and development levels,there are great differences in the automation level of distribution networks in my country.At this stage,there are still many distribution lines with low degree of automation or non-automation.The feeder automation terminals installed on these lines have insufficient coverage,which makes it difficult to accurately locate the feeder automation system.It relies on manual line inspection to check for fault locations.This method cannot quickly and accurately obtain the positioning result,which affects the effect of fault repair and power restoration.First,the principle and implementation scheme of the line fault location method based on feeder-switch power matching are proposed.After the failure of the distribution network,the switch upstream of the fault point will be tripped,resulting in power loss of the feeder.There is a corresponding relationship between the power carried by the trip switch and the power loss of the feeder.According to this principle,using the power information collected in the dis patch automation system and the metering automation system,the feeder loss power and the switch power are predicted and matched by the method of load prediction,so as to obtain the position of the fault trip switch and locate the fault section.Secondly,the calculation method of feeder power loss is proposed.According to the time-varying characteristics of the close relationship between adjacent points of feeder power,IA-LSTM is used to predict and calculate the power loss of the feeder.On this basis,a loss power conversion method for non-synchronized points is proposed to solve the problem that the power sampling points between the feeder and the switch do not coincide.And use the historical load prediction accuracy to set the fault start threshold to solve the problem of lack of alarm information for low and non-automatic distribution lines when a fault occurs.Thirdly,the calculation method of each switching power is proposed.The time-varying characteristics of each switch power of the line,and the correlation between each switch power and external factors and feeder power are analyzed.According to the characteristics of the power carried by different switches,two different prediction methods based on similar days are used for the switching pow er with strong daily periodicity,and then the least squares method is used to fuse similar days to carry out power prediction.For the remaining switches,the power prediction method based on MIV-GRNN is adopted.Tested by historical data,these two power prediction methods have high prediction accuracy for switches with corresponding characteristics.Finally,MATLAB simulation example and Kunming power grid engineering example are used to verify and analyze the proposed fault location method.The analysis results show that the method has good timeliness and effectiveness,and has good engineering practicability.As a fault location method that does not depend on the SCADA system,it is not only suitable for low and non-automatic lines,but also can be use d as an auxiliary method to participate in conventional fault location.This method has been adopted by the "Distribution Network Fault Location System Based on Multiple Incomplete Information" of Kunming Power Grid Dispatching Center in Yunnan Province. |
