Wind Power Prediction Based on Mesoscale Numerical Model and Neural Network Model
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
A mesoscale numerical weather prediction model and the BP neural network forecasting model are used to predict wind farm wind power. By using the WRF numerical model, the meteorological elements from June 2008 to June 2009 of wind are calculated, and results show that the correlation coefficient between forecast and measured wind speed is 072 The forecasts of wind direction, temperature, humidity, and air pressure are also relatively accurate. The BP neural network forecast model of wind power for 40 wind turbines are established, and the influences of the data standardization method and the number of hidden neurons on prediction are analyzed. The results of trial prediction show that the relative RMS error of a single wind generator is 248% to 326%; the correlation coefficient is 045 to 068; and those for the whole wind farm are 215% and 074.
Keywords:
Project Supported:
Clc Number:
You are thevisitors
Supported by:Beijing E-Tiller Technology Development Co., Ltd.
Supported by:Beijing E-Tiller Technology Development Co., Ltd.