Distribution Network Fault Location Based On Artificial Bee Colony Algorithm

Distribution network fault location is used for quickly finding out the faultsections after feeder faults, which is central to the whole fault processing ofdistribution network. It directly affects the correctness and effectiveness of faultisolation and power restoration to non-trouble areas. Rapid fault location laystheoretical basis for achieving self-healing control of the smart grid and has importantpractical significance for reducing the outage time caused by faults and enhancing thereliability of power supply. Learn from the widely used artificial bee colony algorithm,this paper did deeper research on distribution network fault location from the aspectsof analyzing the network topology,application and improvement of the algorithm,processing information distortion or missing and fault location model.This paper proposes a method combining standard bee colony algorithm withfault classification processing for fault location of distribution network, so as to adaptto topology changes of distribution network with distributed power and multi-sourcepower. It uses the criterion of fault classification processing to determine somenon-fault sections, and then locates the fault sections through standard colonyalgorithm, thus greatly improving the efficiency of fault location. The simulationresults demonstrate its effectiveness of the proposed method for fault location. Thestandard bee colony algorithm is likely trapped in local optimal solution and has slowconvergence speed. Therefore, an improved bee colony algorithm is presented,intensifying the standard bee colony algorithm in three aspects: introducing diversitykeeping based on antibody concentration of immune algorithm, modifying the searchequation of leader bees, and adopting the strategy of multi-populationcoevolution.The simulation results of four standard test functions reveal that t heimproved bee colony algorithm is superior to the standard one in computing precision,convergent speed and algorithm stability. The improved bee colony algorithm alsoperforms best in the simulation of a distribution network, compared to the results ofthe standard one and generic algorithm.Then, this paper puts forward the chance-constrained programming model ofsingle fault location, which adopts a switching function taking the influence of falsealarm and missing alarm into consideration, sets the difference between expected andactual state of switches as chance constraints, and establishes the objective function according to the occurrence probability of various faults in distribution network. Theimproved bee colony algorithm based on Monte Carlo simulation is employed to solvethe model. The simulation results show that the model is unable to accurately locatesome fault conditions with a low occurrence probability. So the fault location resultsare further modified by comparing the current amplitude and phase ofpositive-sequence component extracted from electric parameters and this methodsolves the problem successfully.