Research On Single-phase Grounding Fault Line Selection In Distribution Network Based On PSO Optimized Neural Network

With the development of economy and the continuous expansion of the city in China,the service scope of the distribution network is also expanding,its structure is becoming increasingly complex,the impact caused by the failure has also increased.Especially the use of a lot of cables in urban construction causes the ground capacitance and fault current greatly increase.The arc at the fault point is difficult to self-extinguish,which increasing the risk of accident expansion and posing a serious threat to system security.Therefore,when a fault occurs in the system,the fault line should be determined as soon as possible to eliminate the fault and ensure system safety.Although new line selection methods are constantly proposed,most of them have limited application scope,which are either only valid for oone or several fault types,or are affected by neutral grounding mode.A comprehensive and effective comprehensive route selection method needs further study.Firstly,this paper introduces the basic knowledge and analysis the characteristics when a single-phase ground fault occurs.Then,combining the factual circumstances and the need of the paper,a system simulation model is established,which simulates fault types such as arc grounding,high resistance grounding,and transition resistance grounding,and obtains a large amount of fault information data.Then combining wavelet analysis and entropy theory,the fault feature quantities of wavelet energy entropy and wavelet packet energy entropy are extracted.At the same time,the active component and integral parameters of the fault signal are extracted as the fault characteristic quantities.These fault characteristic quantities are used to prepare for the training and learning of the BP neural network in the next.Then the concepts,structure and algorithms of BP network,are analysis.A BP neural network model for picking out the line with wrong established,and an error criterion method in view of error accumulation is proposed.Based on the flaws in the results of line selection,the neural network is proposed optimization.Finally,the principles,processes and characteristics of the PSO algorithm are studied.The optimization of the BP network is completed,and the fault line selection based on the PSO-BP network is realized.By comparing the network before and after optimization,it proves that the line selection method based on PSO-BP network after optimization is more accurate and efficient.This paper has 82 figures,14 tables,77 references.