Accurate Modeling And Simulation Algorithm Of Distribution Network With High Penetratioin Distributed Photovoltaic

In recent years,with the increasing dependence of human society on energy and the increasing depletion of traditional fossil energy,renewable energy,such as photovoltaic power generation,which has the advantages of low-carbon environmental protection and rich resources,has become an important means of energy development strategy in the world.Thanks to the gradual maturity of photovoltaic technology and the encouragement of government policies,small and medium-sized photovoltaic power stations with two-staged structure appear in large numbers and concentrated places in application environments such as roofs of industrial areas and wasteland in rural areas.The high penetration photovoltaic integrate to the distribution network has changed the power flow and dynamic characteristics of the distribution network,making the modern distribution network gradually develop to the active distribution network.Photovoltaic power generation is intermittent and random,large-scale integrate to the distribution network greatly increases the complexity of the grid and the difficulty of operation and maintenance,which has a profound impact on the safety,reliability and economic operation of the grid.Accurate modeling and stable and efficient dynamic simulation are the important basis for the dynamic response analysis and control strategy verification of high PV penetration rate distribution network.It has great significance for the planning and operation control of large-scale active distribution network with random generation characteristics.These distributed integrate photovoltaic power stations form a photovoltaic power station cluster in the local distribution network.Photovoltaic cluster contains many photovoltaic power plants with different parameters,which is a large-scale and complex nonlinear system.How to build a photovoltaic cluster model with low complexity under the premise of ensuring the accuracy of the model is an important challenge at present.In addition,the distribution network with high penetration photovoltaic can also contain other distributed power sources whose dynamic process is different from photovoltaic,which makes it a large-scale nonlinear stiff system,which brings challenges to the efficiency and stability of the traditional numerical integration simulation algorithm.Therefore,this paper takes the distribution network with high penetration distributed photovoltaic as the research object,and carries out the research on the equivalent modeling and simulation algorithm,in order to provide a reliable and effective tool for the research and operation of the modern distribution network.The main contents of this paper are as follows:(1)Dynamic clustering equivalent modeling of distributed photovoltaic power station cluster: in order to analyze the dynamic response of distribution network with high penetration distributed photovoltaic,a new dynamic clustering equivalent modeling method of two-staged photovoltaic power station cluster is proposed.Firstly,after studying the indexes that can describe the dynamic characteristics of the two-staged photovoltaic power station,a Dynamic Affinity Propagation(DAP)clustering algorithm is proposed.Then,the algorithm is used to cluster photovoltaic power stations based on the dynamic characteristics.Finally,the dynamic equivalent model of PV cluster can be obtained by the aggregation of PV station parameters and the simplified equivalent of the network.A photovoltaic cluster with 20two-staged photovoltaic power stations is used to verify the proposed method.The results show that the proposed dynamic equivalent model can accurately reflect the dynamic response characteristics of photovoltaic cluster.In addition,the simplified PV cluster model can greatly reduce the complexity of the model and reduce the simulation time.(2)High precision dynamic equivalent modeling of distributed photovoltaic power station cluster: Based on the limited accuracy of the equivalent model when the clustering parameters are scattered,a high precision dynamic modeling framework of two-staged photovoltaic power station cluster is proposed,that is,deep learning-clustering hybrid modeling framework.It includes clustering based equivalent model and error correction model(ECM).ECM model is constructed by using long and short term memory(LSTM)network,which models the dynamic response error between the existing equivalent model and the detailed model.Finally,based on a practical photovoltaic cluster architecture,a large number of case studies are used to verify the effectiveness of the framework.The simulation results show that this method has the characteristics of low complexity,fast response speed and higher accuracy.(3)Approximate analytical simulation algorithm of distribution network with high penetration distributed photovoltaic: in order to overcome the difficulty of traditional numerical integration algorithm to meet the rapid and stable simulation requirements in the complex nonlinear stiff system environment of active distribution network with high distributed generation penetration,a new iterative method of approximate analytical is proposed.Firstly,the variational iteration method is introduced to obtain the approximate analytical solution of the dynamic components in the two-staged photovoltaic system and synchronous generator system in distribution network with high penetration distributed photovoltaic.Then,the multi-stage mechanism is introduced to apply approximate analytical solution in the continuous time interval,which breaks through the limited convergence domain of the original variational iteration method,so that it can be applied to the simulation of distribution network with high penetration distributed photovoltaic requiring solutions in a larger time span.The proposed simulation method is called multi-stage variational iteration method(MVIM).Finally,simulation examples of different simulation steps and disturbances are carried out in a modified 33 node distribution network and a real large distribution network system.The MVIM proposed in this paper has better stability and greatly improves the simulation speed than the improved Euler method and implicit trapezoid integration method.(4)Research on off-line fast convergence of approximate analytical simulation algorithm: on the one hand,the problem of low efficiency of variational iteration method in off-line calculation is studied.According to several systems of differential equations,the potential link that traditional variational iteration method can speed up greatly in the off-line calculation stage is explained,that is,the obtained approximate analytical solution can be taken into account on the right side of the equation of the current correction function,and then the obtained calculation resources can be used efficiently in a single iteration step,and the first-order differential equation with higher coupling frequency can be calculated preferentially as much as possible The utilization efficiency of the obtained approximate solution is further improved.On the other hand,according to the mathematical model of photovoltaic power station and synchronous generator in the distribution network,the optimization catch-up VIM(OCVIM)formula are derived,which taking into account the known approximate analytical solution and the optimization solution sequence in the current step.The simulation results show that the off-line fast strategy of approximate analytical solution proposed in this paper can not only ensure the high accuracy and stability of on-line simulation,but also greatly accelerate the convergence speed of approximate analytical solution to accurate solution in off-line phase.It is more practical approximate analytical simulation method combined with the previous MVIM for distribution network with high penetration distributed photovoltaic.