Research Of Fusion Fault Line Selection Based On Intelligent Algorithm In Small Current Neutral Grounding System

Once the single-phase grounding fault happened to the small current neutralgrounding system, the fault current is usually quite weak and the system can stillkeep symmetrical. In most cases, the system can still keep running for a while.These are the reasons why the small current neutral grounding system is widely usedin our country’s medium and low voltage distribution network. On the other hand,the fault characteristics aren’t obvious and the traditional fault line selectionmethods usually have limitations. With these reasons, the selection problems are stillunresolved up to now. In recent years, the information fusion technique has madecertain achievements in fault line selection. Expecially with the development ofintelligent algorithms, the area still has a huge development space. In view of theabove backgrounds, this paper carries on the depth research to this problem.Firstly, this paper analyzed the single-phase grounding faults mechanisms ofthe small current neutral grounding system, especially the steady and transient faultcharacteristics. With the simulation model established by MATLAB, theexperiments were carried out to some conventional line selection methods. Withthese works, all methods’ characteristics and adaptability were found. In addition,this paper selected six typical methods as the foundation methods of fusion lineselection.This paper established six fault measurement functions of all kinds of methodsand improved the basic ant colony classification algorithm by increasing the conceptof “contribution attribute node” and adopting a new method to update externalpheromones solubility automatically. With these steps, the algorithm enhanced theglobal search capability. On these bases, picked up the classification rules andproved the effectiveness and correctness of the classification rules. In order tofurther simplify the classification rules and improve the correct rate of fault lines,this paper used the support vector machine for the further processing of the extractedclassification rules and got the final decision function. The parameters of supportvector machine were composed by artificial bee Colony algorithm. Throughcomparing with the ant colony classification algorithm, verified the effect of supportvector machine in the whole algorithm.Finally, by comparing with the BP neural network, verified the effectivenessand superiority of the algorithm used in this paper.