| Summary: | 碩士 === 國立高雄應用科技大學 === 電機工程系 === 98 === This study discusses the evaluation of the small wind turbine used in metropolitan area, and the short-term forcasting of the follow-up power generation capacity. In the beginning, we use Self-Organizing Map Neural Network(SOMNN) to serve as the evaluation tool for the site of wind turbine construction, and categorize many training datum into four kinds through the particular function of SOMNN, then take these four kinds of evaluation as model. However, as to the shot-term power prediction, we take Radial Basis Function Neural Network(RBFNN) as the tool for shot-term power prediction, and use Genetic Algorithms(GA) to improve the learning rate which is set by experiences in the past. Finally, we verify the actual power generation and the prediction power generation of Back propagation Neural Network(BPNN). Furthermore, we use the power predicted by the Neural Network to monitor the actual Power Generation, in order to ensure the operating situation of the wind turbine. At the same time, we can also use the result of prediction for the maintenance schedule of wind turbine.
|
|---|
