Study On Algorithm And Its Application Of Integrating Elementary Tree Transformation And Particle Swarm Optimization(PSO) On Distribution Network Reconfiguration(DNR)

With the improvement of requirements of power supply reliability and power quality for power customers, State Grid Corporation committed to take measures to narrow the average annual outage time between urban areas and rural areas this year, besides lowering the loss of active power of distribution network(DN), and solving the problem of low voltage at rural areas. As the end of the power system, DN is directly connected to the power customers, and it is dorminated by overhead lines, and there are more branch lines, which leads to the high rate of active power loss and failure, besides poor voltage quality. Distribution network reconfiguration(DNR) can improve the above problems, and without the need for upgrading the existing DN, which brings significant economic and social benefit. However, the character of "closed loop design, open loop operation" of DN, besides a lot of switch combination produces a large number of solutions without meeting the requirements of topology radial constraint during the process of DNR which are needed to be judged and rejected.With the development of intelligent technology, there is larger extension space for DNR by using intelligent optimization algorithm. This paper adopts the particle swarm optimization(PSO) algorithm, improving the global convergence ability of the algorithm by dynamically adjusting the inertia weight and learning factors. The simplification of initial DN topology can simplify the process of calculation. The spanning tree strategy of the elementary tree transformation is introduced, first looking for a basic tree from the simplified network, generating different trees through exchanging the tree branch and link branch which correspond to different radial network structure. Combining different trees with PSO algorithm makes the search space of DNR limit to the feasible solution space which meets the requirement of radial topology, without the need for calibration, to improve the convergence efficiency. Initializing the position and velocity of swarm by encoding, assuring different groups of open and closed circuits after the exchange between tree branch and link branch by decoding,different running status of network are obtained which can update individual and global extreme value, finally the global optimal solution is got.Choosing the minimum active power loss as objective function, taking standard DN as an example to make a DNR, the simulation results show that, the convergence character of PSO algorithm with dynamic inertia weight and learning factor is more stable than traditional PSO algorithm. Comparing with other algorithms, integrating elementary tree transformation and PSO algorithm can effectively reduce the active power losses, improving voltage level, more stable global convergence, higher computation efficiency.Finally, the algorithm is used to solve the problems of Smart Grid Research Institute of DNR and fault restoration reconfiguration. The results show that, DNR can significantly reduce the active power loss, reducing the load balanced coefficient, improving voltage level, and fault restoration reconfiguration can correctly locate the position, isolating fault, effectively restoring load, which makes its engineering application value desired.