Research On Multi-criterion Of Small Current Grounding System Single-phase Fault Line Detection Based On BP Neural Network

In China’s 10~35kV low-voltage distribution network,single-phase ground fault accounts for 80%~90% of the total number of all faults.The neutral point of such distribution network is usually operated in non-effective grounding mode(ie,a small current grounding system).When a single-phase ground fault occurs in this small current grounding system,the short-circuit fault current is very weak,making the fault information difficult to capture due to the high impedance of the short circuit loop.As the structure of the distribution network becomes more and more complex,and branches feeders have become much more,so the requirements for the accuracy and rapidity of fault line detection methods become more stringent.In this paper,based on the research and application of the fault line detection method by many researchers at home and abroad,combined with BP neural network and genetic algorithm to achieve multi-criterion fusion fault line detection method,in order to improve the accuracy and speed of fault line detection.First of all,this paper introduces the domestic and foreign research status in the field of fault line detection,sums up the problems existing in the current research,and finds the breakthrough point of the subject.At the same time,the theoretical analysis of the steady-state and transient feature of the fault,the use of digital signal processing method to extract fault feature,builda line detection criteria based on the fault feature amount,to lay the foundation for the establishment of multiple criteria fusion fault line detection model.Secondly,the BP neural network is briefly introduced,and the fusion of multiple fault line detection criteria is achieved by using its powerful nonlinear fitting ability and self-learning ability.Then,a small current grounding system simulation model was built in Matlab/Simulink to simulate various types of fault conditions,and the fault line detection model based on BP neural network was simulated and verified,confirming its feasibility and superiority.Finally,for the problem that the traditional BP neural network randomly generates the initial weights and thresholds of the network,which leads to its slow convergence speed and easy to fall into the local dead zone,the genetic algorithm is added for optimization.Through the simulation,the network performance of the traditional BP neural network and the optimized by the genetic algorithm is compared and analyzed.It is verified that the BP neural network fault line detection method based on the genetic algorithm optimization has the advantages of high accuracy and high speed of the fault line detection.