Long term wind speed estimation for a randomly selected time interval by using artificial neural networks, Amasra, Turkey


Nogay H. S., Akıncı T. Ç.

ENERGY EDUCATION SCIENCE AND TECHNOLOGY PART A-ENERGY SCIENCE AND RESEARCH, cilt.28, sa.2, ss.759-772, 2012 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 28 Sayı: 2
  • Basım Tarihi: 2012
  • Dergi Adı: ENERGY EDUCATION SCIENCE AND TECHNOLOGY PART A-ENERGY SCIENCE AND RESEARCH
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED)
  • Sayfa Sayıları: ss.759-772
  • Anahtar Kelimeler: Long term wind speed estimation, Artificial Neural Network, Forecasting, ENERGY, DENSITY
  • Bursa Uludağ Üniversitesi Adresli: Hayır

Özet

In this study, a new ANN estimation model has been developed in order to estimate wind speed in the long term. The data used in the model developed for this study was provided from the wind station in the province of Amasra, Turkey. In the analysis, average wind speed per hour data for the period 2002-April 2009 were used. Wind speed measurement range is 10 minutes. As the input of the ANN model, wind speeds per hour during the day, year and month numerical values were used. As the output of the said model, randomly specified wind speed at 16:00 hours during the day was used. This means the estimation of the wind speed at 16:00 hours on any date by using the data between the years 2002 - 2009 was conducted. While in recent studies in the literature, estimation of wind speed in the short term was performed by using time series, times series was also used in this study, however, unlike other studies, our study is based on the estimation of wind speed at any randomly specified time or moment. By this investigation, estimations on any date and time specified in the future can be done. The objective of this study is to estimate the wind speed at only a single time with only the ANN model with a very high estimation percentage. In this study, the wind speed at a specified time has been estimated with a very high percentage with only the ANN model and error percentage realized on a minimum level. Quite successful results were achieved in terms of both arrangement of the data set and realization of the estimation of a single minute.