Research On Reactive Power Resource Allocation Of Grid Forming/Following Preventing Multi-timescale Voltage Fluctuation

The proposed novel power system has accelerated the process of replacing traditional power generation methods with renewable energy generation applications.Among which,distributed power generation is an effective means for large-scale development and utilization of renewable energy.However,voltage regulation in distributed power generation systems is difficult owing to its randomness and intermittency characteristics,which cause the voltage of the distribution network to significantly exceed the limits.Moreover,modern power electronic equipment has the characteristics of controllable operation,diverse functions,and specialized/combined hybrid usage,which present new approaches for reactive power compensation.Accordingly,this study conducted an in-depth assessment on the grid forming/following reactive power resource allocation to suppress voltage fluctuation and regulated the voltage from a multi-timescale perspective.The specific contents include the following:Considering the uncertainties in photovoltaic output and power consumption characteristics in the distribution network,a stochastic modeling method for the disturbances on the power distribution network driven by power consumption data was proposed,which paves the way for establishing uncertainty operation scenarios of power distribution networks.The power consumption characteristics of the station area were described based on Gaussian mixture clustering models,while the photovoltaic output characteristics and states were described based on the cloud state indicators and Beta distribution.A behavioral state time sequence relationship was established according to the Markov chains principle,sequential Monte Carlo was used to randomly sample the duration of behavior state,and a random model of the power consumption behavior state was established for the disturbance source.Latin hypercube sampling was used to sample the power consumption characteristics of the disturbance source.Moreover,the power consumption data of the disturbance source was generated by considering the probability density function.The forward/backward power flow algorithm was used to analyze the voltage distribution characteristics of the distribution network and evaluate the risk of voltage overrun.The accuracy and effectiveness of the proposed modeling strategy are verified by IEEE 33 node distribution network.Considering the time-varying characteristics of reactive power compensation in the distribution network,a multi-timescale cooperative control strategy for reactive power resources considering the auxiliary governance of grid following photovoltaic inverters was proposed.The reactive power compensation mechanisms for the grid following/forming power electronic inverters were analyzed.The shunt capacitor bank(SCB),static var generator(SVG),and photovoltaic inverters were classified according to the equipment tracking performance.Subsequently,the corresponding relationship between reactive power equipment and multi-timescale optimization was established according to the time-varying characteristics.A multi-timescale reactive power optimization model of the distribution network structure/network equipment was established aiming to minimize the voltage deviation of the entire network.The particle swarm algorithm was used to optimize the output power curve of the reactive power equipment with different response levels;finally,it achieved the multi-timescale reactive power optimization of the grid forming/following equipment.The proposed strategy is compared with other methods to verify the rationality and effectiveness of the proposed strategy.According to the grid forming/following control characteristics of photovoltaic inverters,a multi-timescale cooperative control strategy for reactive power resources considering the auxiliary governance of grid forming/following distributed photovoltaic(DPV)cluster was proposed.A DC topology considering the characteristics of string photovoltaic inverters was proposed to enable the DPV clusters to generate a certain number of free-state inverters and flexibly switch between grid following or grid forming reactive power optimization as needed.A two-stage reactive power optimization model was established,in which the first stage considered the minimum voltage deviation of the entire network as the objective.The second stage considered the maximum number of free-state inverters as the main objective and optimal DPV cluster operation performance as the auxiliary objective.Based on the objectives in the second stage,a strategy to prevent the inverter power from exceeding the limit and prevent a frequent switching action was proposed.Moreover,a distributed cluster control decision-making scheme was proposed.Finally,the multi-timescale reactive power optimization of the distribution network was achieved using both DPV cluster and special equipment.A simple DPV cluster control system is composed of three photovoltaic array simulators and three photovoltaic inverters to verify the rationality and effectiveness of the proposed control strategy.The multitimescale reactive power optimization strategy is compared with other methods by using IEEE33 node distribution network to verify the rationality and effectiveness.Based on the difference in the cost of reactive power equipment and time-varying characteristics of voltage regulation,a cooperative optimal allocation strategy for grid forming/following reactive power resources based on distribution network partition was proposed.The method to establish the stochastic model of the disturbance source was used to establish the operation scenario of the distribution network.In addition,the number of free-state inverters in the DPV cluster was analyzed considering the time series of photovoltaic output,which provided the basis for selecting the equipment types for reactive power optimization.Then,the node sensitivity index was established.The reactive power optimization area of the distribution network was divided based on the community discovery algorithm,and the average sensitivity index was used to select the regional dominant governing node.The chance constrained programming was used to address the inequality constraints of distribution network operation scenarios.A two-layer optimal configuration model ensuring minimum investment cost and voltage deviation over the entire network was established.Furthermore,the particle swarm algorithm was used to optimize the capacity of special reactive power equipment.The rationality and effectiveness of the proposed strategy are verified by comparing different schemes.