Research On Prediction Of Output Power On Megawatt-class Wind Turbine

Wind power is a clean and renewable energy with aboundant and widely distributed. As with great development potential, wind power can be commercialized and scaled in short time. But the difficulty of power dispatching is increased because of the random, intermittent and volatility of wind power. So, the forcast of output power timely and effectively can reduce the running costs and spinning backups, thus the effects of the power grid will be mitigated, the utilization of wind power will be improved. Especially, the studies of forcasting on the output power of million watt wind power system have an importantly reality significance.The relationships between wind speed characters of wind power plant and output power are analysed based on1.5MW wind power generator system in this thesis. The accurate forcasting of wind speed in wind power plant is the premise and key of the output power of wind power unit.In order to get a more accurate forcasting of wind speed, the fuzzy neural network of wind speed forcasting which is the combination of fuzzy system and neural network for the characters of wind power plant is used. In this method, historical wind speed, wind direction, rotor speed of wind power unit and pitch angel are inputs to build the wind speed model of fuzzy neural network, and the simulation experiment is done. Then the results of the fuzzy neural network of wind speed forcasting and the results of BP (Back Propagation) neural network of wind speed forcast are compared. The simulation results show that the fuzzy neural network of wind speed forcasting can get a better result. It can lay the foundation for output power of wind power unit.The output power forcasting of time series based on wavelet neural network, the output power forcasting of grey neural network and the output power forcasting based on least squares support vertor machine regression of output are proposed in this thesis. The forcasting of output power of MW wind power units is done by using tempature, pressure, relative humidity, wind speed direction of wind plant and the forcasted speed as inputs and the simulation experiments are done. In order to get a weighted average result, a proper weigh is selected for the above three methods restively. The combination forcasting model is established to eliminating the large deviation of single forecast method, at the same time, the accuracy of the output power forcasting is improved. The simulation results indicated that the combination forcast is with a high accuracy.