Fault Line Selection Based On Random Matrix And Fault Section Location Based On Improved Bat Algorithm For Distribution Network

With the development of smart grid and the increasing demand of users for the reliability of power supply,the fault handling ability of power grid needs further improvement.The fault condition of distribution network is complex and can lead to the power outage.The rapid and accurate fault line selection and fault section location after fault need to be paid more and more attention and further research.In China,non-direct ground neutral system is mainly adapted to distribution network and single-phase ground fault is the most common fault in the distribution network.Because of the weak fault current,the fault feature is not obvious.So the single-phase ground fault line selection and the fault section location are difficult.At present,the technology of fault line selection of non-direct ground neutral system has made great progress.With the deep research of the data mining technology,it is a good way to use data analysis technology to detect the fault line.The single-phase ground fault section location method of non-direct ground neutral system is in the research stage and lacks practical application experience.So,further research is needed.Using the recording data of faulty current,a method of fault line selection in distribution network based on random matrix theory is studied in this paper.The original data matrix is obtained by using the amplitude of the current before and after the fault.The original data matrix is partitioned and translated.Then the state data matrix is generated.The distribution of eigenvalue vectors of the state data matrix is analyzed by random matrix theory.For the non-fault feeder,its eigenvalue distribution is in accordance with the single ring theorem of random matrix theory,most of which lie in a particular circle.For the fault feeder,its eigenvalue distribution appears "contraction".The Hausdorff distance algorithm is used to analyze the similarity of the eigenvalue vectors of different feeder lines,and the average Hausdorff distance of the eigenvalue vectors of the state data matrix of each feeder is obtained.The fault line selection criterion of the distribution network is established.The distribution network model with 5 different line types is built in MATLAB/Simulink.Simulation results under different fault scenarios verify that the proposed method can accurately select the fault feeder under different fault locations,fault types and transition resistances and is suitable for ungrounded system and resonance ground systemIn this paper,an improved bat algorithm is applied to locate the fault section in distribution network.The transient zero-mode current energy value of each point of the feeder is constructed.And according to the transient zero-mode current feature of each measuring point on the feeder,a suitable fuzzy membership function is constructed to obtain the contrast of transient zero-mode current energy,which is used as the input of the bat algorithm.A bat algorithm based on fuzzy inference mechanism is used to solve the problem of fault section location.The membership function of the fitness function value is established to make the bat amend its speed factor according to the current fitness function value in each optimization process.By this way,the speed is controlled in real time.The modified algorithm improves the defect that it is easy to fall into local optimum and enhances the ability of optimization.The results show that the method in this paper is suitable for single phase grounding fault under different transition resistance,fault locations and initial fault angle.The modified bat algorithm avoids falling into the local solution too quickly and it can better locate the fault section than the classical bat algorithm.