| In recent years,with the increase of DC power-using devices,the economic cost of power systems is increased through the large number of converters used.Therefore,the traditional AC power grid is difficult to meet the development needs of society.The energy crisis is alleviated with the advent of wind-photovoltaic-storage DC microgrid,improved environmental problems and decreased the use of converters.DC microgrid is accepted distributed renewable energy generation systems such as wind and photovoltaic,energy storage units,electric vehicles and other DC power loads more efficiently and reliably.It is important to promote energy saving and emission reduction and achieve sustainable energy development.Due to the small damping of DC lines and the poor stability of renewable energy output,the occurrence of faults is aggravated the harm to the wind-photovoltaic-storage DC microgrid.Therefore,fast and accurate detection and location of DC lines is one of the key technologies to ensure the stable operation of wind-photovoltaic-storage DC microgrid.Fault detection and positioning are the key to ensuring the stable operation of DC micro grids.At this stage,there are fewer fault detection and positioning research methods at the landscape storage DC microgrid,and the length of power grid structure and transmission and distribution DC lines is very different from other types of DC transmission systems.Based on the structure and output characteristics of the wind-photovoltaic-storage-DC microgrid,the structure and distributed power model are analyzed,the common electrical fault types and fault characteristics are summarized,and the fault characteristics of inter-pole short-circuit faults and single-pole grounding faults are analyzed in this thesis from the theoretical aspect with emphasis on the change of current and voltage signals after faults in microgrids with different topologies.Aiming at the problem of current signal feature selection for DC line fault detection and location in wind-photovoltaic-storage-DC microgrids,a method for selecting eigenvalues based on the maximum weighted correlation index based on empirical modal decomposition is proposed.The DC current measured at the transmitting end is decomposed to obtain multiple eigenmode functions.The maximum weighted correlation index method is used to calculate the two selected eigenmode functions with the highest sensitivity,and nine statistical methods are applied to the eigenmode functions.The function generates fault characteristic quantities.This method reduced the dependence on the original signal and preserves the best features of the fault detection data.Aiming at the problems of lower fault detection accuracy,longer fault detection time,and difficulty in locating different types of faults,this thesis proposed a method for fault detection and localization of DC lines in wind-photovoltaic-storage-DC microgrids,the decision tree and the gradient boosting decision tree is combined to detect and locate fault.The decision tree is used to detect the faults of the DC line and distinguish different fault types,and the gradient boosting decision tree is used to locate the lines of the interpole short circuit and the ground fault type respectively.The method is used to detect and locate faults on DC lines of windstorage DC microgrid quickly and effectively regardless of the fault resistor,fault onset moment and fault distance.In this thesis,the wind-photovoltaic-storage-DC microgrid model is constructed by Matlab/Simulink,and the efficacy of the proposed method is tested using Python.By randomly setting the fault conditions（fault resistor,fault onset moment and fault distance）,it has enhanced the realism of the wind-photovoltaic-storage-DC microgrid model simulation,and the effectiveness of results is validated through the proposed method. |
