Fault Line Detection In Small Current Grounding System Based On Improved Artificial Bee Colony Algorithm Evolving Neural Network

The fault line detection in small current grounding system is still a difficult problem in power system field,with the expansion and complexity of domestic distribution network,and the development of traditional power grid to smart grid,the solution of the problem becomes more important.In all kinds of small current grounding fault line selection methods,most of them are single criterion method based on steady information or transient information.The automatic line selection device based on these methods is not suitable for complex distribution network,and the accuracy of line selection is low.In this thesis,a multi-criterion fault selection scheme is proposed through the analysis and comparison of existing line selection method.In this method,zero sequence active component,high harmonic synthesis quantity,transient energy and neural network theory are used for the fault line selection.The scheme makes up the shortness of the single criterion,completes the fusion of intelligent population algorithm and neural network theory,and improves the fitness and accuracy to complex distribution network.Firstly,the single-phase ground fault simulation model of small current grounding system is constructed in Matlab/simulink,by changing the relevant parameters,a lot of simulation is carried out for single-phase ground fault in all kinds of situations,on the basis of simulation,the fault eigenvalue of each criterion is collected and a large number of fault feature data are recorded.Secondly,aiming at the shortage of BP neural network,a novel artificial bee colony algorithm is introduced to optimize the neural network in the thesis,and the levy flight walk mechanism is introduced to improve the global optimization ability of the artificial bee colony algorithm.Finally,the fault feature data is input into the optimized neural network,and the fault line is identified successfully.For the same test sample data,the basic BP neural network and improved artificial bee colony algorithm evolving BP neural network is tested respectively,through comparison and analysis of the simulation,fault line selection method in this thesis which based improved artificial bee colony algorithm evolving BP neural network has higher accuracy and better effect,the method has theoretical research significance and engineering practical value.