|
With the increasingly wind power penetration, the impacting of grid-connected wind farm to the power grid will be increasingly serious, which maybe damage the power system operation of economic, security, stability. Wind is the driving force of wind power, but wind energy has strong randomness and volatility, making wind power penetration sharp fluctuations, the major impact of grid-connected wind farm to the system include: frequency stability, power angle stability, voltage stability, harmonics, voltage fluctuation and flicker, net loss and the trend of the distribution, back-up cost, schedule planning, system reliability and other aspects. A good wind power forecasting results can allow operators to make dispatch plans, to arrange for the combination of unit and back-up system, and so on, as a result, the impact of grid-connected wind farm to the power grid will be reduced.On the one hand, the difficulty of the wind speed prediction lies in: wind can be affected by many factors, including the geographical and meteorological, geographic factors such as topography, landforms, latitude and others, meteorological factors such as on the temperature, wind levels, humidity and so on, changing of each factor may lead to the change in wind speed, these facts make less regularity of daily variation of wind speed, forecasting results can not make satisfactory; On the other hand, researchers can analyze the relevance between wind speed and various factors, and extract useful information by the relevance of wind speed with factors, which are useful to wind speed forecasting. In this paper, the methods, including intuitive method and statistics method are used to analysis the relevance of wind speed with the main meteorological information. Intuitive method is as following: selecting wind speed and the relevant weather information in a continuous period of the date, according to the size of wind speed value, observing the changing trend as the wind speed increases; statistical method is suing the concept of relevant factor in statistics, calculating the relevant factor of wind speed with weather information and anglicizing the relevant nature (a positive correlation or negative correlation).In this paper, the short-term combination forecasting based on fuzzy clustering technology is proposed by the combination of BP neural network model and fuzzy clustering technology. Firstly, based on the correlation analysis of wind speed with the main meteorological information, some meteorological factors are extracted by the their relevant degree with wind speed, these extracted meteorological factors are made the sample indicators of fuzzy cluster; Secondly, since the relevant degree of each factor with wind speed is different, this paper impose a weight on each factor and gives the mathematical method of calculating index weight, furthermore, the accurate clustering results is unnecessary in the wind speed forecast, so this paper also improves the traditional fuzzy clustering technique to s apply them conveniently. Finally, the wind speed forecasting results are obtained by BP neural network model which the inputs are composed of part of the meteorological information and the history wind speed values, futhermaore, wind power forecasting results are obtained also by the power curve of the wind generator.According to the measured datas from two wind fram in Shanghai, by comparing the foreacasting results which are obtained only by BP neural network model and the forecasting results obtained by the method in the paper, the results indicate that the prediction accuracy is increased by 5% at least, so it is concluded that this method is fully effective and practical.
|