Preprocessing And Recognition Method Research Of Large Scale Wind Turbines Based On SCADA Data

With the rapid development of wind power industry, the capacity of wind turbine increasing,single unit cost is becoming more and more higher, the wind machine in the event of failure,will increase the wind turbine manufacturers and maintenance costs of wind farm operator,meanwhile the various types of wind turbine accidents also serious negative influence on the healthy development of the wind power industry. Scholars has made a lot of work based on wind turbine SCADA data for the operation of the wind turbine condition monitoring and fault diagnosis, because of the SCADA data contains a large number of random information inconvenience to processing, and the existing research is mainly concentrated in the theory and method of exploration phase. Based on a 2 MW direct drive wind turbine as the object and the original wind turbine SCADA data in wind farm, a data preprocessing method is established and constructed the evaluation strategy, analyzes the random of wind influence wind turbine power output, then build the wind turbine performance degradation model and provide reference for wind turbine design and control. The main research contents of this thesis are as follows:In order to obtain more explicit physical information from the SCADA data and judge the operating state of wind turbines better, the mean value method, least square method and non-parametric method(kernel density-mean value method) are used to pre-process SCADA data in wind farm. Three evaluation indexes for pre-processing algorithm are presented,including: 1. the consistency of physical characteristics; 2. the robustness of the sampling time; 3. the robustness of the sampling frequency. The quantitative calculation formulas of the evaluation indexes are designed to evaluate the effect of various kinds of pre-processing algorithms. Based on the SCADA data in full working condition are pre-processed, operating characteristics of wind turbines are analysed, including the relationship between the wind speed and output power, rotational speed of the wind rotor and the wind energy utilization coefficient.The random fluctuation of wind is the basic reason that causes the output power fluctuation of wind turbines. Based on the relation model of “wind—power of wind turbines” and giving full consideration to the influences of wind speed and wind direction changes to the power fluctuation of wind turbines,the wind speed fluctuation coefficient, the wind direction fluctuation coefficient and the comprehensive influence factor are defined. Then one dimensional evaluation model of wind speed fluctuation to power fluctuation, wind direction fluctuation to power fluctuation and two dimensional evaluation model of wind speed andwind direction fluctuation to power fluctuation are presented. The operation process of wind turbines is divided into three regions in this paper: the constant region of wind energy capture coefficient, the transition region and the constant power region. The characteristics of power fluctuation are analyzed for each region.Put forward the evaluation index of wind turbine performance degradation and its mechanism analyzed, established wind energy utilization coefficient, power fluctuation characteristic, engine vibration characteristic, temperature fluctuation features a description of the form. Using single wind turbine history state longitudinal comparison and many wind turbine of horizontal comparison, using the method of contrasting and weighted average method to evaluate the wind turbine performance degradation, judge the degradation degree of the wind turbines.