Study Of Short-term Prediction Of Wind Power

Wind power, the most commercial value renewable energy, is developing rapidly in therecent years. However, due to the intermittence and randomness, wind power will causeproblems such as power quality and stability when grid-connected, which greatly restrictingthe increase of wind farm capacity and grid-connected ratio. Wind power predictiontechnology can solve the problems and contribute to generation plan and fan repair, which canalso improve the quality and economic benefits of wind power generation.Wind power prediction has been researched for a long time; commercial predictionsystem is mature abroad and arises interiorly. However, due to the board area, complicatedterrain and incomplete NWP in China, study of wind power prediction needs to be improved;the main works of this paper are as follow:(1) Analyze the distribution and change regulation of wind speed and wind direction; identifythe use of different numerical weather prediction, provide a basis for modeling.(2) Study the general technology in wind power prediction, select support vector machine asthe modeling method after the comparison of different model characteristics, which hasgood performance in small sample modeling.(3) Taking into account the lack of data processing methods in wind power prediction, thispaper puts forward a new method based on similarity and sequential time clustering,which considers the daily similarity and sequential information of wind speed, improvesthe similarity and reliability of training data. Compared with other clustering methods,similarity and sequential time clustering has a higher prediction accuracy andgeneralization ability.(4) A wind power prediction system is developed based on the research of this paper, featuresinclude regular data collection, automatic forecast and report and remote monitoring,which is running steadily in some wind farms and achieved satisfactory results.