Research On Optimal Control Strategy Of Hierarchical And Partition Of Urban Distribution Network For New Energy Consumption

In the context of energy transformation,large-scale distributed renewable energy is incorporated into the power grid,resulting in significant changes in the operation characteristics of the power grid.In the case of incomplete meteorological information,it is difficult to effectively perceive the operation situation.The quantity of power supply and load is huge,the types are diverse,its randomness and volatility are increasing,and the number of optimization variables is increasing rapidly.The traditional dispatching control mode can not meet the operation requirements of urban distribution network in the future any more.This paper studies the optimal regulation technology of urban distribution network for new energy consumption.The main work and achievements are as follows:Firstly,aiming at the incomplete or missing meteorological information of distributed photovoltaic,considering the time-shift characteristics of meteorological information,a mechanism-data hybrid driving power prediction model of distributed photovoltaic is proposed.The meteorological data provided by adjacent public meteorological stations or centralized photovoltaic power stations equipped with numerical weather prediction(NWP)are used to solve the problem of meteorological data source.A mechanism-data hybrid driven power prediction model is established.For the mechanism driven model,according to the historical power curve of the target station and the irradiance data in NWP,the optimal time shift of meteorological data relative to the photovoltaic site is analyzed and the meteorological data is adjusted.For the data-driven model,multiple time patterns are extracted through onedimensional convolution,and the attention mechanism is introduced to overcome the problem of long-distance dependence of RNN,that is,the memory of RNN decreases with the increase of its input length.Further,by integrating the advantages of mechanism-driven model and datadriven model,the prediction accuracy of the model is improved.The effectiveness of the method is verified by German distributed photovoltaic data and meteorological information data.Secondly,aiming at the structural goal of partition decoupling control and the functional goal of energy balance,a two-stage dynamic partition method of urban distribution network is proposed.In the first stage,considering the time variability of the supply and load as well as the stability of the partition,the historical data of the node net load is discretized according to the season to obtain the comprehensive electrical distance expectation.Taking the comprehensive electrical distance expectation as the measurement,the partition is carried out by the improved fast search and find of density peaks algorithm.In the second stage,on the basis of the first stage,fine-tuning of the edge nodes of the partition is carried out with the aim of power balance based on the forecast results of photovoltaic and load power.The effectiveness and rapidity of the method are verified by the extended IEEE33 node distribution network.Thirdly,aiming at day-ahead scheduling of distribution network,a multi-objective dayahead optimal scheduling strategy based on mixed integer second-order cone programming is proposed to make full use of multivariate controllable resources in urban distribution network.Considering the adjustable resources such as static var compensator,on-load tap changer,capacitor bank,interruptible load,energy storage system and photovoltaic in urban distribution network,through the convex relaxation of power flow equation and the accurate linearization of distribution transformer,this paper establishes a multi-objective model to improve the consumption rate of clean energy,reduce network loss,improve voltage level and improve user satisfaction,so as to realize the coordination and optimization of active and reactive power.The effectiveness of the method is verified by extending the example of IEEE33 node distribution network.Forthly,aiming at the intra day regulation of distribution network,a "rolling and real-time" multi-time scale partition distributed intra-day optimal scheduling framework is constructed.In the rolling optimization stage,the decomposition coordination optimization model is constructed based on the partition of urban distribution network,and the model is solved by the alternating direction method of multipliers(ADMM)algorithm.In the real-time regulation stage,based on the feeder control error of each zone,each zone further realizes the error correction according to the established control mode,balances the system power fluctuation and forms a closed-loop control.The effectiveness of the distributed dispatching framework is verified by extending IEEE33 distribution network simulation.