Optimal Voltage Control For Active Distribution Systems

With the increasing of distributed generations(DG),the operation and control of active distribution systems(ADS)is faced with enormous challenges.Voltage control is an important issue in active distribution systems.Voltage of buses in active distribution systems exhibits obvious spatial and temporal characteristics that voltage of buses is not usually reduced from the root buses to the end buses and greatly affected by DG operation modes,which makes the coordination of the voltage control devices more complicated.Partitioning and pattern recognition can help ADS accommodate these characteristics.Partitioning based on voltage sensitivity can effectively enhance the voltage control performance of the voltage regulation devices and reduce the control cost.Energy storage systems(ESS)configured on the basis of partitioning information can ensure sufficient voltage regulation devices in the partitioning.Mode recognition and prediction for output power of photovoltaic(PV)systems can help decouple models within the dispatching days and speed up calculations.Research works on the optimal voltage control strategy for ADS have important theoretical significance and application values for rationally coordinating voltage regulation devices and ensuring voltage quality.This paper focuses on the research of optimal voltage control for ADS.The main work and research results are as follows:(1)Partitioning for ADS and ESS configuration.In order to reflect the impact of the change of power flow,node distance is defined based on the voltage sensitivity and the improved clustering by fast search and find of density peaks is used to obtain the real-time dynamic partitioning information.To ensure sufficient voltage control devices in each partitioning of the ADS,it is necessary to acquire the union set of three-phase partitioning in a period of time and configure ESS in each partitioning.Then,ESS locations can be obtained by using the weighted index including degree,closeness and betweenness in the graph theory.Besides,an optimization model considering the voltage deviation,electricity price and installation cost is established to obtain reasonable capacities for ESS.Simulations designed in this paper verify performance of the proposed partitioning method and ESS configuration strategy.(2)Pattern recognition and prediction for output power of PV system.Categorical generative adversarial networks driven by monthly historical output power data are trained to obtain a generator with capability of capturing the characteristics of the output power and a discriminator with classification characteristics.Then,pattern for output power of PV systems can be divided into sunny,overcast,cloudy and rainy modes.Combined with weather information given by the weather forecast,multi-mode prediction curve for output power of PV systems can be obtained by trained generator in a joint way.The prediction curve can be updated by the curve that has the largest correlation coefficient with the measured output power curve.Simulations are designed to verify the proposed pattern recognition and prediction methods.(3)Optimal voltage control strategy for ADS.To speed up calculation of the optimal models,fuzzy clustering algorithm is firstly used to cluster the load modes.Then,the dispatching days can be decoupled according to day modes of PV systems and night modes of load.RDP(Ramer Douglas Peucker)algorithm is used to select feature points of PV curves and load curves to further reduce the number of time points in dispatching days.To reduce control cost of voltage control devices,real-time dynamic partitioning information is introduced by modifying the voltage reference values in the conventional voltage objective function.To optimize operation of voltage control devices,different control strategies are made for each mode.Then,an optimal model accounting for OLTC(On Load Tap Changer)constraints,ESS constraints and voltage constraints is formulated that can be solved by multi-objective particle swarm optimization.Simulations are designed to verify the effectiveness of the proposed optimal voltage control strategy.