| At present,with the vigorous promotion and use of clean energy,the development of offshore wind farms is becoming more and more rapid.The capacity of offshore wind turbines is increasing and the overall scale is also growing.With the development of offshore wind farms,the internal high-frequency fast transient overvoltages caused by frequent operation or failure of electrical equipment in its collection system have become more and more serious.The longterm accumulation of overvoltages will cause damage to the insulation protection of electrical equipment.In severe cases,it will even damage the equipment and affect the safe and stable operation of the system,which brings huge economic losses.In order to prevent and solve the problems caused by overvoltages,the classification and identification of different types of fast transient overvoltages in offshore wind farms is of great significance to the overall insulation coordination and protection setting of offshore wind farms.In order to analyze the characteristics of fast transient overvoltages in offshore wind farms,this paper simplifies the topology and equipment parameters of an offshore wind farm in Guangdong Province,and builds a laboratory platform to obtain four types of fast transient overvoltages.However,considering the conditions,cost,and safety of the field test,the types and numbers of experimental samples obtained are limited,and the feasibility of researching only by collecting a small amount of experimental data is not convincing.Therefore,according to the structure and equipment parameters of the laboratory platform,a simulation platform that can reproduce the high-frequency fast transient characteristics of this type of overvoltages is built on the PSCAD/EMTDC software to expand the scale and type of experimental samples and improve the scientificity and effectiveness of this research.Then based on the obtained simulation and experimental samples,the characteristics of the fast transient overvoltages in offshore wind farms can be more comprehensively investigated,which lays the foundation for the identification of various fast transient overvoltages in offshore wind farms.This paper first proposes a multi-scale mathematical morphological signal feature extraction method.By constructing a new structure element,the multi-scale mathematical morphological decomposition method is employed to extract the high and low frequency components of fast transient overvoltages,which are constructed as the time domain characteristic quantities for identifying different types of internal fast transient overvoltages.Then based on the constructed high-frequency feature and the high-low frequency energy ratio feature,a support vector machine classifier model is employed to classify various types of internal fast transient overvoltages.In addition,to make up for some shortcomings of multi-scale mathematical morphology,this paper also proposes a sparse representation feature extraction method.First,a redundant discrete cosine dictionary and an identity matrix dictionary are used as the harmonic dictionary and the impulsive dictionary,respectively.Then the alternating direction multiplier method is used to update multiple iterations,and the harmonic components and impulsive components of the original overvoltage signals are extracted based on the dictionaries.Then combined with the impulsive components and the original signals,the impulsive feature and the energy ratio of the impulsive components to the original signals are constructed.Then a support vector machine classification model is employed to identify different types of fast transient overvoltages in offshore wind farms.The simulation and experimental results show that compared with the conventional wavelet algorithm,the proposed two methods have a more obvious distinction of the features,better feature extraction capabilities.They can accurately identify various types of fast transient overvoltages in offshore wind farms,which lay the foundation for the protection setting and insulation coordination of electrical equipment in offshore wind farms. |