Fault Location Technology Of Distribution Network Based On Traveling Wave Theory And Convolution Neural Network

The development of the economy has led to the demand for electricity Ssupplly.which has brought about the complexity and scale of the power grid,and brings new challenges to the safe and stable operation of the power grid.Distribution network as the terminal skeleton of the power grid,its importance has become increasingly prominent,the quality of the distribution network power supply requirements are also increasing.In the case of single-phase fault,the distribution system with neutral point is directly connected with the reliability.It can be used for 1-2h,so it is widely used in low-voltage distribution network in China.Although in the event of a single-phase fault allow with trouble running in a period of time,but it a single-phase fault with trouble running in a long time,may lead to equipment insulation damage,endanger the safety ot power supply.Therefore,in order to avoid the single-phase ground fault caused by the above problems,we need to design a fault location method to quickly troubleshooting.The rapid,accurate and reliable fault distribution of the distribution network,and the timely elimination of fault to recovery power supply,to improve the reliability of power supply is of great significance.According to related statistics,single-phase ground tailure is the most frequent type of failure occurred in the distribution network.Due to the non-neutral grounding method,the tault current is small.Therefore,the traditional distance measurement method based on tault steady state is limited.However,the fault location based on fault transient can effectively avoid the above defects.The traveling wave ranging method has the characteristics of simple structure,and is not affected by the grounding mode and other factors.For a multi-branch radial distribution network,it is only necessary to install a single-ended traveling wave ranging method at the bus at a distance is more advantage than the two-way traveling wave method.However,the single-ended traveling wave method still has the difficulty of recognizing wavefront,and the wave velocity is not determined.At the same time,with the rise of artificial intelligence technology in recent years,researchers began to try to apply artificial intelligence technology to the distribution system.Because artificial intelligence technology has strong feature extraction,classification and recognition ability for image and sound waves,artificial intelligence technology can be used to classify fault waveforms in distribution network to determine the fault location.This paper combines the traveling wave ranging with artificial intelligence technology,to further improve the fault location technology of distribution network.In this paper,combined the theory of traveling wave ranging which not affected by neutral point grounding and other factors,and the convolution neural network with excellent image classification and recognition ability,this paper proposes a distribution power system fault location technology which combine the convolution neural network and traveling wave theory,and through ATP-EMTP software simulation to verify the feasibility of the above method.At the same time,compared with the traditional double-end traveling wave ranging method,the advantages of the proposed method are illustrated.In conclusion,the feasibility and correctness of the fault location method of distribution network based on traveling wave theory and convolution neural network are verified by theoretical analysis,simulation and physical simulation platform.