| In recent years,the photovoltaic grid connected power generation industry has expanded rapidly.However,with the large-scale photovoltaic grid connected,the problem of islanding becomes more and more prominent.Therefore,the detection and identification of islanding and the research of islanding protection are playing an important role.At present,the islanding detection technology in the field of grid connected inverter power generation system is mainly divided into the following two types:passive and active.As we all know,many PV inverters are often incorporated into the power grid,so the islanding detection and identification of multi inverter connected to the power grid has also attracted more and more attention.Therefore,this paper studies how to distinguish the islanding state and non islanding state more effectively and better on the basis of multi inverter grid connection of distributed photovoltaic power generation.Firstly,the paper analyzes the islanding detection method in the case of single inverter connected to the grid simply and concisely;secondly,it analyzes the detection performance of the active frequency offset method and the active frequency offset method with positive feedback which are widely used in the case of multi inverter connected to the grid theoretically;finally,the active islanding detection method in the case of multi inverter connected to the grid will reduce To identify the islanding problem accurately,a pure passive detection method based on the variational mode decomposition and neural network is proposed.For the new method,firstly,the basic theory of VSM and neural network and the detection principle of the new method are introduced;secondly,the extraction and formation of the characteristic vector of islanding voltage signal based on VSM and multi-scale entropy processing are studied in detail,as well as the training process of neural network for islanding state judgment,so that the islanding state of multi inverter grid connected is realized Finally,the proposed islanding detection method is applied to simulate five different states of the system running under the condition of multi inverter grid connection,and then the voltage signals under each condition are extracted,processed and judged,finally the detection effect is shown by data statistics.The simulation results show that the multi inverter grid connected islanding method based on the variational mode decomposition and neural network has high detection accuracy and relatively stable simulation data.It can effectively distinguish the islanding and non islanding two different operation states,which has relative practical and reference significance. |