Equivalent Modeling Of Distributed Photovoltaic Power Station Clusters And Dynamic Simulation Of Distribution Network

In recent years,the development of distributed renewable energy generation has been rapid,large-scale access to the distribution network has become an inevitable trend.However,a large number of intermittent and random distributed generations(DGs)have a significant impact on the safe,reliable and economical operation of the power grid.Distributed photovoltaic is the distributed power supply with the largest number of access in the distribution network.Dynamic simulation of distribution network with distributed photovoltaic cluster is an important means to analyze the interaction between photovoltaic power generation and power grid.However,the large number of power stations in the photovoltaic cluster leads to high model order and long simulation time.At the same time,the multi-time-scale characteristics of dynamic processes are obvious,and the degree of nonlinearity is enhanced.In view of the above problems,this paper takes active distribution networks(ADNs)with large-scale access of distributed photovoltaic clusters as the research object,and conducts research from two aspects of dynamic modeling and simulation algorithm of photovoltaic power generation clusters:(1)The single-machine modeling method of photovoltaic power generation system suitable for dynamic simulation of distribution network was studied.On the basis of the detailed model of electromagnetic transient of photovoltaic power generation system,the switching period average model of DC/DC chopper and the quasi-steady model of grid-connected inverter are established,and the simplification and order reduction of the detailed model are realized.Set up the first order differential algebraic equations(DAEs)photovoltaic dynamic model,easy interface and network Algebraic equation is calculated.The simulation results show that the proposed dynamic photovoltaic model has the advantages of both accuracy and computational efficiency.(2)The dynamic equivalent modeling method of distributed photovoltaic cluster was studied.Firstly,the spectral clustering algorithm is used to carry out multi-machine equivalent modeling for distributed photovoltaic cluster.Taking the typical dynamic parameters of photovoltaic power generation system as the clustering index,a spectral clustering algorithm is introduced to cluster the index.Secondly,the parameter aggregation and network parameter equivalence calculation of photovoltaic power station group after clustering are carried out,and the multi-machine equivalence model of photovoltaic cluster is established.Aiming at the accuracy of the clustering equivalent model,the Deep Belief Network(DBN)is used to carry out nonlinear fitting of the external characteristics of the detailed model of photovoltaic cluster,and the nonmechanism dynamic model of photovoltaic cluster is established.Based on the improved IEEE33 distribution system,the dynamic characteristics of the detailed model,the clustering equivalent model and the nonmechanism model are compared,and the results show that the non-mechanism model has higher precision than the clustering equivalent model.(3)The dynamic simulation algorithm of distributed active photovoltaic power distribution network with high density is studied.Firstly,the numerical characteristics and numerical stability of classical numerical integration algorithm are analyzed in detail.Combined with the active power distribution network is presented for the problem of strong rigidity,strong nonlinear characteristics,puts forward two-stage semiimplicit Runge-Kutta(2s-SIRK)numerical integration algorithm to solve the dynamic simulation model of differential equations(DEs).Among them,a parameter optimization strategy and an Adaptive Jacobian Matrix Update(AJMU)strategy are proposed to improve the numerical performance of 2s-SIRK by comprehensively considering the numerical stability,computational efficiency and computational accuracy.Based on the improved IEEE33 distribution system and the actual distribution system in Jinzhai,Anhui province,the effectiveness of the proposed algorithm is verified.