Research On The Optimization Algorithm For Distribution Network Service Restoration

China’s demand for power is growing and the electric power industry continues todevelop. The power grid automation evolves to a higher level and the distributionautomation is an important part. When the distribution network’s failure occurs,completing the identification and isolation of fault and restoration of power failure arearapidly is an essential task of distribution automation.According to the actual operation of the distribution network, researching theefficient and practical algorithm and forming restoration software have great significancefor quickly restoring lost power supply of non-fault area which lost power. Distributedgeneration technology is widely used in the distribution network, and it has a greatimpact on the traditional distribution network. Therefore, the study of distributionnetwork service restoration may not ignore the impact of DG.This paper conducts in-depth research on the distribution network servicerestoration. According to the features of the distribution network’s structure andoperation and the distribution network service restoration’s requirements for fast andreliability, a three-phase power flow algorithm based back/forward sweep method isproposed as a basis for subsequent restoration. The distribution network servicerestoration will be divided into two parts. Firstly, for the DG in the area out of power, usedramatic islanding scheme which use online dynamic search to form the best islandbased on the offline data pre-generated. Then, regard the islands as scheduling load andparticipate in the remaining distribution network service restoration. Two recoveryalgorithms are proposed. One is the improved BPSO. It provides the particledisplacement update rules to ensure distribution network’s radial operationalrequirements and make some improvement for its defect that is easy to fall into the“premature”. Another is branch exchange-Particle Swarm Optimization which combinesthe Branch exchange’s fast and efficient and the global optimization of PSO. The particledisplacement is taken as contact switch and the branch exchange id used to determinewhich section switch to open for the formation of ring. The programming test is madeand the comparison of calculation result of IEEE33-node distribution network showsthat the branch exchange-Particle Swarm Optimization has the higher computationalefficiency. Finally, an example with DG is established to verify the reliability of theresults of this algorithm and simulation test of practical demonstration line verifies thepracticality of the algorithm.Considering the characteristic of distribution network and the influence of DG，thealgorithm can obtain restoration the results in short time. With the growing access of DG, this algorithm has very high practical value. Applied to the distribution automation,it can rapidly form the fault recovery scheme and assist dispatchers to make decision.