TY - GEN
T1 - Generation of simulated wind data using an intelligent algorithm
AU - Weissbach, R.
AU - Wang, Wen-li
AU - Hodge, B. M.
AU - Tang, M. H.
AU - Sonnenmeier, J.
PY - 2014/11/21
Y1 - 2014/11/21
N2 - Wind energy is becoming an important renewable resource for a residence with no access to grid-based power. However, its variable and uncertain nature requires the use of energy storage to ensure high reliability of power to the residential loads. Therefore, variations in the wind profile need to be considered more accurately to predict energy storage requirements. Various models aim to develop simulated sets of wind data to meet this purpose, including Numerical Weather Prediction (NWP), statistical time series approaches such as Auto-Regressive Moving Average (ARMA) models, Artificial Neural Networks (ANN), and Markov models. Unfortunately, generating statistically significant wind data from these models for a single location has proven to be difficult. In this paper, an intelligent algorithm is developed to create and impose simulated wind data to tackle variations at a single site. It first incorporates the Markov approach to identify the general trend of the measured data and ensure fidelity of the Probability Density Function (PDF). A learning methodology is then employed to ensure the trend also satisfies seasonal and diurnal constraints.
AB - Wind energy is becoming an important renewable resource for a residence with no access to grid-based power. However, its variable and uncertain nature requires the use of energy storage to ensure high reliability of power to the residential loads. Therefore, variations in the wind profile need to be considered more accurately to predict energy storage requirements. Various models aim to develop simulated sets of wind data to meet this purpose, including Numerical Weather Prediction (NWP), statistical time series approaches such as Auto-Regressive Moving Average (ARMA) models, Artificial Neural Networks (ANN), and Markov models. Unfortunately, generating statistically significant wind data from these models for a single location has proven to be difficult. In this paper, an intelligent algorithm is developed to create and impose simulated wind data to tackle variations at a single site. It first incorporates the Markov approach to identify the general trend of the measured data and ensure fidelity of the Probability Density Function (PDF). A learning methodology is then employed to ensure the trend also satisfies seasonal and diurnal constraints.
UR - http://www.scopus.com/inward/record.url?scp=84918493906&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84918493906&partnerID=8YFLogxK
U2 - 10.1109/NAPS.2014.6965405
DO - 10.1109/NAPS.2014.6965405
M3 - Conference contribution
AN - SCOPUS:84918493906
T3 - 2014 North American Power Symposium, NAPS 2014
BT - 2014 North American Power Symposium, NAPS 2014
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2014 North American Power Symposium, NAPS 2014
Y2 - 7 September 2014 through 9 September 2014
ER -