Power distribution system is at the end of the power system, it connects with consumers directly, so it is important to make sure the power distribution system is in both safe and economic operation. The main purpose of the reactive optimization about distribution network is to achieve the balance of reactive power, reduce active power loss, improve the quality of power supply and ensure the system in safe and stable operation according to distribute the reactive power reasonably and effectively.As the distribution network is complex, when dealing with the reactive power optimization directly, it may lead to some problems, such as large amount calculation and more iterations. So before optimizing it, we can make the power distribution network into several independent areas, which is also called distribution network partition, then optimizing each area at the same time, because the subsystem is smaller than original system, the iterations will decrease when achieving the optimal value. But in traditional partition methods, regional partition just takes natural quality into account, ignoring the electrical characteristics; Graph partition method is not consistent with practical operation of distribution network; and clustering partition cannot guarantee that there is enough reactive power in every area. Therefore, it is necessary to study the distribution network partitioning.Firstly introduce the present situation and the significance about reactive power optimization of distribution network in detail, and make a contrast about all kinds of optimization algorithms on both disadvantages and advantages, meanwhile, build a mathematical model of reactive power optimization; Then aiming at the defects in current distribution network partition, this paper processes a new method:based on the voltage-reactive sensitivity method, define electrical distance matrix, then regard the central pivot node as the initial node, calculate the electrical distance between each node and central pivot node, On the basis of the electrical distance, emphasizing on considering whether the node voltage meeting the demand of power quality, at the same time, put regional attribute into considering, then re-partition by calculating the electrical distance again for those nodes which do not meet the requirements. After that, with the help of the objective function and the expert knowledge to merge or disaggregate the initial partition, so the final partition forms. Finally, the particle swarm optimization based on improved inertia weight is applied to the distribution network reactive optimization, and writing programs by MATLAB, take IEEE-33 and PG&E69 node systems for example,according to compare the result between partition and non-partition, We can draw a conclusion that it is feasible and superior to optimize the distribution network after partition. |