Font Size: a A A

Research On Ultra-short Term Wind Power Forecasting

Posted on:2017-03-19Degree:MasterType:Thesis
Country:ChinaCandidate:S Y ChenFull Text:PDF
GTID:2272330482982382Subject:Power system and its automation
Abstract/Summary:PDF Full Text Request
Due to the intermittence and fluctuation of wind power, the operation of large-scale wind power integration can do harm to the safety and stability of the power system, even wind power curtailment which lead to reduce the utilization rate of wind energy resources. So forecasting the wind power is very important. It can coordinate the relationship between conventional energy and wind energy for power system dispatching department. Therefore, this thesis studies on the question that how to improve the precision of the ultra-short term wind power forecasting.In order to improve the precision of the ultra-short term wind power forecasting, in view of the problem that traditional BP neural network relies a lot on samples, through in-depth analysis of entropy method and expert scoring method, similar days are selected by combining above two methods in this thesis. It has optimized training samples and excluded the incorrect parts of the sample. Then the model of ultra-short term wind power forecasting is established by utilizing the BP neural network algorithm to forecast wind power and reduce prediction error effectively.According to the problem that point forecasting cannot get high prediction with the large wind speed fluctuation. This thesis focuses on interval forecasting of power based on the optimization of the training samples. First, the data is divided into 5 sections according to the fluctuation of power. The probability density curve in each section is obtained by using the kernel density estimation. Then find the confidence interval and fluctuation range of the forecast day. It can enhance precision of ultra-short term wind power forecasting further.By optimizing data from a wind farm in Jiangsu, the direct and indirect methods are applied to finish point forecasting and interval forecasting. The conclusion proves that methods mentioned in this thesis do improve the precision of ultra-short term wind power forecasting effectively.
Keywords/Search Tags:ultra-short term wind power forecasting, BP neural network, entropy weight, similar days, interval forecast
PDF Full Text Request
Related items