Research On Wind Power Prediction

Wind power prediction technology and its support system plays an important role in electric power and energy balance, grid dispatching operation and control, provinces electric power transaction, wind farm production, maintenance and repair. Foreign wind power prediction technology study began in the nineteen eighties, with thorough research and wide application, the development of technology and application of product become more and more mature. China’s relevant application development began in2008, after several years of rapid development, There are about100research institutes or developers, but problems and deficiencies after operation still exist,especially in the accuracy, reliability and adaptability to different wind farm. These forecast system product can not meet the actual production needs, especially in high altitude, complex terrain and centralized development of large-scale wind power bases, the performance of forecast products is more poor. This paper is based on anemometer tower network and each wind turbine information, studies wind power forecasting technology by using statistical methods and combining of physical and statistical methods, these results have been applied to the Gansu wind power forecast system construction to improve the forecasting accuracy and reliability and obtained apparent effect, which has important academic value and practical engineering value. The paper mainly includes the following several aspects of the work:1. This paper introduced the development of wind power at home and abroad, the status quo and development trend of wind power forecast, numerical weather prediction in MM5V3mode and WRF mode process, characteristics and functional modules, summarized the characteristics and basic situation of China and Gansu wind power development, analyzed the problems of wind power base in Jiuquan. Studied the numerical forecast products such as artificial neural network, studied the wind power prediction technique based on physical method and parameter selection from the aspects of atmospheric motion, local effect, wind speed and direction, carried out the super short-term wind power forecasting based on the correction technique development.2. This paper completed data assimilation at home based on the observation, according to weather climate characteristic in Jiuquan area, improved meso-scale regional WRF model of wind element forecast. In the meantime, completed the WRF mode business optimization and downscaling method in technology development, realized wind farm’s element forecast in Jiuquan about38prediction point (34wind farms,4towers of the wind observation point) on wind speed, wind direction, wind power, air pressure, temperature and other factors, the prediction time is72hours, interval forecast is15minutes, forecast level is10,30,50,70meter. Daily forecast based on the meteorological data assimilation is in8:00and20:00. The paper based on the historical observation data analysis established the wind element of short-impending prediction equations and developed short impending prediction products by using the statistical method and mathematical methods.3. According to the Gansu wind farm’s climate characteristics and historical data, a set of norms prediction system was developed by using statistical method and combining methods of statistical and physical for building predictive models. The system can satisfy the needs of power dispatching and standardize operation of the power system, realized real-time monitoring the information of23wind towers which basically covers Gansu area,42wind farms and more than4000Wind turbines, achieved wind power short-term prediction of a single wind farm, wind farm groups and the whole province. Most of wind power’s short-term forecasting error is less than15%, the error of total power grid is less than11%.Experience certificated two prediction model be a good applicability for dispatching department and provide important reference for preparation of power plan.4. This paper took information of real-time wind velocity, wind direction, each wind turbine and substation as input, combined with wind power prediction results of each prediction point after the numerical weather forecast were descended scale, realized super short-term wind power prediction in Gansu province by Hybrid prediction method based on the time series, Most of wind farm’s super short-term power prediction is less than12%, the error of total power grid is less than9%.which can meet the needs of real-time scheduling.