Research On Islanding Detection Technology Of Distributed Generation Cluster

As the world 's energy demand grows year by year,the rapid consumption of primary energy and the environmental degradation it brings are becoming increasingly serious.The government has issued a financial subsidy for grid-connected photovoltaics and a priority grid-connection policy.The number of photovoltaic inverters has increased dramatically and has the characteristics of clusters.In this case,new requirements and challenges are proposed for the original islanding detection method for a single inverter.For the situation of dilution effect and power quality interference in the grid connection of distributed power generation clusters,it is of great significance for the system to detect islands quickly and accurately.In this paper,the research on island detection technology of distributed power generation cluster is carried out.The research contents are as follows:(1)Describe the structure of the distributed generation cluster scenario and its island mechanism,analyze the principles,advantages and disadvantages of the existing island detection methods,and point out the difference between the method of this paper and the existing methods and the comparison of the existing island detection methods.Advantage.Build the simulation model of the corresponding scene,describe the details of the model building process,and formulate the formula for the change of the voltage and frequency of the grid connection point under different conditions of power matching.(2)In view of the shortcomings of the stand-alone power exchange island method in this scenario,the grid-connected transformer no-load loss is considered as the criterion and the reactive power difference is used as the island criterion.Combined with the passive detection method,a distributed power generation cluster-based power exchange is proposed Method,and verify its correctness from both simulation and experiment perspectives.(3)For the shortcomings of the power exchange method at the grid connection point,it is necessary to switch on the large capacitor and may affect the power quality.An island detection method based on wavelet packet entropy is proposed.This method only needs to extract the voltage waveform data from the grid connection point for wavelet packet entropy processing,and then select the appropriate Judgingthe threshold can accurately judge the island state of the system.Finally,in order to verify the anti-interference performance of the method,additional switching capacitors and resistors,voltage sags and sags,and harmonic injection are added.The simulation results show that the method has certain Anti-interference.(4)For the feature extraction of wavelet packet entropy and the selection of appropriate thresholds,it is not easy to propose an island detection method based on Stacked Sparse Auto-Encoder.Simulation shows that this method does not require manual feature extraction when the network structure and parameters are properly selected,and there is disturbance It can also accurately detect the island operating conditions of the distributed power generation cluster.