Research On Real-time Simulation And Modeling Of Large-scale Photovoltaic Power Generation System Based On Clustering Algorithm

With the rapid development of the photovoltaic(PV)industry,the grid connection of large-scale PV power generation systems has increased the penetration rate of power grids.A large amount of power electronic components and non-linear loads have been added to the power system,which has seriously polluted the power system.Therefore,it is especially important to model and simulate large-scale PV power generation systems.However,if the large-scale PV power generation system is modeled in detail,there will be problems such as large simulation scale and long simulation time.Therefore,it is necessary to construct a simplified equivalent model of large-scale PV power generation system with both accuracy and simplification.In this paper,the PV power generation unit is firstly modeled based on the RT-LAB real-time simulation platform.For the AC grid and DC transmission line,the CPU-based real-time simulation system is used in subtle steps.And for the inverter with high frequency switch,the FPGA-based highspeed real-time simulation system is used in nanosecond steps.The real-time simulation waveform is compared with the actual inverter experimental waveform to verify the accuracy of the detailed model.Then,the composition of the large-scale PV power generation system is described and its model is presented.This paper constructs the inverter equivalent simplified model by using the controlled source instead of the voltage source converter,constructs the synchronous generator set model to simulate the frequency fluctuation of the power system,and constructs the ZIP load model to simulate the actual load characteristics.Then,based on the real power plant parameters,the real-time simulation is verified,and the automatic voltage control fixed reactive power control algorithm is implemented based on the built model.In order to further simplify the large-scale PV power generation system,the illumination intensity,temperature and real-time active power which can characterize the characteristics of the PV power generation unit are used as clustering indicators.Based on the fuzzy C-means clustering algorithm,the large-scale PV power generation system is classified.At the same time,the equivalent parameter calculation method is used to aggregate a class of PV power generation units into a PV power generation unit,and the equivalent model is constructed.The RT-LAB real-time simulation is used to verify the effectiveness of the equivalent model under fault and disturbance conditions.Finally,the traditional fuzzy C-means clustering algorithm is improved.Before using the fuzzy Cmeans clustering algorithm,the Canopy clustering algorithm is used for preprocessing,which improves the problem that the fuzzy C-means clustering method relies on the initial clustering center point.Considering the complex network topology of large-scale PV power generation system,the equivalent line impedance of each photovoltaic power generation unit from the grid-connected point is used as a clustering index.Based on RT-LAB real-time simulation platform,the accuracy of the clustering equivalent model and the rationality of the proposed clustering method are verified.