Load profile partitioning and dynamic reactive power optimization

The optimal reactive power control scheme changes throughout a day due to the time-varying load demands. In order to reduce switching operations of control devices, the load demand profile is usually decomposed into several sequential segments, in which the load can be regarded approximately invariable. Two new load profile partitioning methods are proposed in this paper. One of them uses an optimization method, where the objective function is inherited from typical clustering problems, and the other uses a heuristic iteration method to partition the load profile by evaluating the maximum load deviation (MLD). Furthermore, a simplified dynamic reactive power optimization model is presented based on load curve partitioning. Its objective function considers not only the energy loss but also the voltage quality for different loads. Based on the energy loss, voltage level and switching times of control devices, three load profile partitioning methods are compared in detail. The Ward-Hale 6 buses system is used to illustrate the effectiveness of the proposed load curve partitioning and reactive power optimization methods.