Research Of Oscillation Identification And Modal Analysis Based On Large Data Of Wind Farm

With the adjustment and transformation of the energy structure,the penetration of new energy sources represented by wind power in the power grid has been continuously increasing.Under the influence of many kinds of external uncertainties,the phenomena of oscillation and harmonics occur simultaneously during the operation process of wind power system connected to grid.It is an important method to study the operation of wind farms by data analysis.Given the current situation that many wind farms have thousands of wind generating sets in practice,it is difficult to conduct a precise mathematical modeling of wind farms.Even if the mathematical model of wind farms can be established,it is possible to encounter dimensionality disaster problem which can be hard to be solved.In the actual operation of the wind farms,the system will produce a large amount of data every moment,meanwhile,these multi-source,heterogeneous and complex data also contain valuable information related to the wind farms.This paper probes into the methods suitable for the big data in wind farms,so as to lay a solid foundation for the identification of the oscillation type in wind farms.First of all,the current situation of wind power development is introduced,the interaction between grid and wind farms and harmonics are concerned,and the commonly used methods for oscillation identification and modal analysis are introduced.Secondly,according to the current situation of research on the data in wind farms,the stratification method of data in wind farm is researched,and the data acquisition scheme in wind farm is discussed.Based on the large doubly-fed wind farms data acquisition system designed in this paper,a modular data acquisition device is developed,and a system framework used for acquisition with smallscale data and processing was built.Thirdly,based on the established model of doublyfed wind power system connected to grid and combining FFT method,HHT transform,Prony algorithm and other methods,the phenomenon of oscillation was explored and the oscillation mode is classified in detail.Finally,the harmonic problem is studied by using improved FFT and k-medoids clustering algorithm.And based on k-means clustering algorithm and apriori algorithm,the influence of cluster of wind speed and voltage on the oscillation modal in wind farm is studied.Wind power is predicted based on longterm and short-term memory networks,which can be used to predict the problems of the interaction between grid and wind farms.The research in this paper reveals the phenomena of oscillation and harmonics,which can be used to guide practice.