Research On Fault Location For Medium Voltage Distribution Network With DGs Based On Neural Network

The interconnection of a large number of DGs has completely changed the traditional single power supply and radiant network topology of distribution network,making it develop into a new type of power distribution system with two-way flow,which leads to the failure of the traditional fault location algorithm of distribution network,and bring new challenges to various fault location algorithms based on distribution automation.This paper studies the present situation of fault location of distribution network with DGs,proposes a new fault location method of distribution network with DGs based on neural network.And on this basis,a new fault location method of distribution network with DG based on two-layer multi-agent system is proposed.Specific work contents are as follows:(1)Considering the influence of ranging error and multiple branches of distribution network,this paper proposes to input fault characteristics of each power point into RBF neural network for fault ranging,and obtain multiple fault ranging results based on different power points.Considering the influence of the transition resistance,the transition resistance is added to the output characteristic vector of the RBF neural network for fault location.Taking 33-bus distribution network of IEEE as an example,the simulation results show that the distribution power electric parameters is introduced into fault location input layer of the RBF neural network model can significantly improve the ranging precision of the model,the short circuit transition resistance is introduced into fault location output layer of the RBF neural network model can effectively weaken the negative influence of the transition resistance on fault location results and significantly improve the fault location accuracy of the neural network model(2)In view of the increasingly obvious distributed characteristics of multi-branch and multi-node complex distribution network,this paper proposes a two-layer multi-agent framework for fault location of the main agent and the main-branch agent.The main-branch agent is used to upload fault information,and the main agent is used to centrally analyze the ranging results.Considering the influence of distance error and multi-branch of distribution network,a new method for fault location of distribution network with DGs based on the main agent and main-branch agent multi-agent system is proposed.This method uses the information correction of adjacent line segments to solve the impact of measurement errors on fault identification near the dividing point of adjacent line segments,uses the current measurement information at the first end of branch lines to eliminate pseudo fault line segments,and performs fault location isolation by main branch agent calibration,thus solving the problem of uncertain fault location of multi-branch parallel lines.Taking 33-bus distribution network of IEEE as an example,the simulation results show that fault location based on the main agent and main-branch agent two-layer multi-agent system can correctly locate the fault section and the specific location of the fault.(3)In order to verify the applicability of the model and method proposed in this paper,the actual distribution network of Qihe country is taken as an example for simulation analysis.The simulation results show that the fault ranging method and fault line segment positioning method proposed in this paper are also applicable to the actual power distribution system with large scale,the ranging accuracy meets the requirements,and the positioning results are accurate.