Studies On Methods Of Network Partitioning For Reactive Power Voltage Control Based On Community Mining

Large-scale interconnectedness of power system improves the utilization ratio of resources and it also brings complex reactive power voltage control problem. Voltage running within the allowable range through the monitoring of node voltage and the control of reactive power source, is not only beneficial to ensure the voltage quality of the system operation, but also can improve the security of the system.Hierarchical voltage control is an important means to improve the voltage quality of power system. Network partitioning for reactive power voltage control is one of the key issues in the hierarchical voltage control method. Community structure in complex network has a great similarity with the partition for reactive power voltage control. This paper researches network partitioning methods based on community mining thought in community structure theory, and its mainly contents are as follows:(1) A network partitioning method based on betweenness of line and community mining is proposed. Firstly, taking into account the quasi-steady physical response of PV nodes, reactive voltage sensitivity between generator and load nodes is calculated to obtain the initial pre-partitions on the premise that generator nodes have high voltage control capability to the load nodes inside the partition. Secondly, betweenness of line is proposed to ensure the reactive coupling of partition in the merger process, which is calculated based on reactive power flow tracing. Then, weak coupling lines are determined by the betweenness of lines. The final partition is obtained through merging the adjacent partitions without weak coupling link lines. Finally, partition reactive power balance degree index is given to check the reactive power balance of partitioning result. The simulation results show that the proposed method can ensure reactive power coupling and balance of partitions on the premise that generator nodes have high voltage control capability to the load nodes inside the partition.(2) A network partitioning method based on reactive power source-load betweenness and community mining is proposed. Considering that reactive power from reactive source will affect the voltage of the nodes on the transmission path between reactive power source-load nodes, reactive power source-load betweenness is proposed to measure the reactive power coupling between reactive power source-load nodes. Reactive power coupling modularity function for network partitioning is determined based on the betweenness and community structure in complex network theory. For the shortcoming that branch with high betweenness open enumeration states are too single in the search for community structure using traditional community mining method so that it can't obtain the optimal partition result, optimization model for reactive power network partitioning is set up with the goal of maximizing reactive power coupling modularity and with the connectivity and static reactive power balance constraints. The model is solved by genetic algorithm. The simulation results show that the proposed method from the perspective of reactive power transmission can ensure the global optimality of partition, at the same time, in the case of small load growth, the partition has a certain stability.