A methodology to forecast wind power production 24 hours ahead is developed using a
hybrid approach of an artificial neural network (ANN) and numerical weather prediction (NWP).
The methodology is simple and designed to be applicable to any wind farm on the globe, using
publicly available NWP data and basic historical power production data from wind farm.
Notably, no historical wind data from on-farm sensors is required as the 0 hour forecast data is
used to train the ANN. The results are encouraging, with a root-mean-square-error of 0.2267 for
a 24 hour ahead forecast, corresponding to a forecast error standard deviation of 0.23 per unit. / Thesis (M.S.)--Wichita State University, College of Engineering, Dept. of Electrical Engineering and Computer Science.
| Identifer | oai:union.ndltd.org:WICHITA/oai:soar.wichita.edu:10057/5031 |
| Date | 07 1900 |
| Creators | Pfeifer, Mark B. |
| Contributors | Jewell, Ward T. |
| Publisher | Wichita State University |
| Source Sets | Wichita State University |
| Language | en_US |
| Detected Language | English |
| Type | Thesis |
| Format | x, 265 p. |
| Rights | © Copyright 2011 by Mark B. Pfeifer. All rights reserved |
Page generated in 0.0014 seconds