Research On Load Reduction Control Of Wind Turbine And Power Allocation Of Wind Farm Based On Support Vector Machine

In recent years, due to the growing influence of environmental pollution caused by traditional energy sources, countries have stepped up research efforts for new energy sources. At present, the wind energy, as one of the best renewable energy, has attracted the people’s attention. With the development of wind power technology, the unit capacity of wind turbine is increasing which improves the efficiency of wind turbine power generation, but presents new difficulties for the wind turbine control technology.For the large scale wind turbine, larger blades mean bigger loads. The traditional independent pitch blade technology can effectively reduce fatigue loads, however, it cannot effectively attenuate the limit loads caused by the gust wind. In addition, there is no great way to generate power in accordance with the needs of the grid for large-scale wind farm. The output power of wind turbines is still determined by the method of maximum power point tracking, resulting in a waste of electricity and reducing the whole wind farm efficiency.In this paper, load reduction and power scheduling issues were studied, and the main contents include the following four aspects:(1) This paper proposes a new method based on SVM for processing natural wind. GH Bladed was used for modeling natural wind and outputting the actually measured characteristics of the wind turbine. Through the sample under different wind conditions, the SVM model, including the SVC and SVR, is built for determining the gust wind and estimating the wind speed. Based on models available, the results of wind conditions classification and wind speed estimates can be got.(2) This paper proposes a novel composite load reduction control strategy aiming at the wind turbine load reduction. By using the SVC model, this strategy first determines the current wind conditions, and then uses appropriate control strategy depending on different conditions. In this way, traditional independent pitch control and tower vibration compensation control under gust wind are effectively combined together. The effectiveness of the proposed method is validated by GH Bladed.(3) On the issue of power scheduling in the wind farm, this paper uses SVR model to estimate the wind speed through the easily measured factor of the wind turbine. This method can solve the problem of the error and latency of measuring wind speed on the top of the nacelle. According to the estimated wind speed, the power required for grid is allocated to the wind turbine under different wind speeds. And the desire power of each turbine will be determined according to the actual situation.(4) In addition, this paper focused on the power tracking controller based on robust adaptive algorithm to maximize the wind energy factor Cp. It is shown that good control performance is ensured by the proposed method under unknown degeneration of system parameters. Stability of the algorithm is proved by the Lyapunov stability.This work was supported by the National Key Basic Research Program of China (Grant No.2012CB215202), the National High Technology Research and Development Program of China (Grant No.2012AA052302) and the National Natural Science Foundation of China (Grant No.51207007).