Research On Identification Method On Converter Faults In Doubly Fed Induction Generator Based On Power Quality And Knowledge Graph

With the continuous advancement of energy production and consumption revolution of China,new energy power generation including wind power has developed rapidly.Under the guidance of carbon peaking and carbon neutrality goals,the proportion of wind power generation in the grid has been increasing consistently.For wind power generation,the fault of the converter of doubly fed induction generator usually have great concealment,which makes it difficult to find and eliminate the fault at the first time and affects the safety and stability of the whole system.When a fault occurs in the converter of doubly fed induction generator,the output voltage and output current fluctuate,resulting in power quality disturbance.By monitoring and analyzing the features of the power quality disturbances,the fault of the converter can be realized.Based on power quality disturbance caused by converter fault and its typical characteristics,combined with knowledge graph technology,this thesis proposes a fault identification method of doubly fed induction generator converter based on power quality and knowledge graph.The main innovative works in this thesis are as follows:(1)An adaptive power quality disturbance feature parameter extraction method is proposed.Instantaneous negative sequence component method is selected to extract three-phase unbalance feature parameters,and short-time Fourier transform is selected to extract harmonic feature parameters.Fault start time is obtained by analyzing extracted three-phase unbalance feature parameters to determine the start time of short-time Fourier transform window,and short-time Fourier transform is carried out to obtain harmonic feature parameters.The method has high accuracy in extracting fault start time,and is able to locate short-time Fourier transform window according to fault start time,which proves to have better adaptive ability.(2)According to the extracted power quality disturbance feature parameters,a fault identification method of doubly fed induction generator converter based on power quality disturbance feature parameters is proposed.First,on the basis of extracted fault start time,analyze the monitoring parameters which extract the start time of power quality disturbance,and preliminarily identify the converter fault;Then,appropriate harmonic feature parameters are selected,and combined with machine learning algorithm,the converter fault is further identified,and eventually realize the accurate identification of various types of converter faults.(3)By adopting knowledge graph technology to the fault identification method of converter,the knowledge graph for fault identification of doubly fed induction generator converter is established.First,the overall framework of knowledge graph is built.Then,extracted from system topology and disturbance feature parameters,the corresponding entities and relationships are generated,and the identification queries of different faults are established,so as to realize the systematic and rapid identification of doubly fed induction generator converter faults.(4)Aiming at avoiding the influence of external faults of doubly fed induction generator system on the fault identification method,the converter fault identification method is further improved.First,differentiation method for external fault is designed.Then,the framework of knowledge graph is extended,and the queries for external fault differentiation are established to accurately distinguish the converter fault from the short circuit fault in external power grid.The simulation results indicate that the fault identification method of doubly fed induction generator converter in the thesis can realize the corresponding fault identification with high accuracy and good robustness in the system of doubly fed induction generator paralleled with power grid.Meanwhile,the method can also accurately distinguish converter fault from short circuit fault in external power grid,which has good generalization ability and practical application ability.