Wind speed prediction using non-gaussian model based on Kumaraswamy distribution

ABSTRACT

Wind power is a clean source of energy that not only helps in meeting the growing electricity demand but can also play a profound role in transforming the global energy distribution to mitigate the impact of climate change. Accurate forecasting of wind speed is a crucial factor of power system management. The present endeavor focuses on the development of a non-gaussian framework, Kumaraswamy seasonal autoregressive moving average (KSARMA), to model and forecast the wind speed data of 12 locations across the Indian subcontinent. The monthly wind speed dataset during the period 2003 to 2019 has been utilized to perform the analysis. Diagnostic analysis, Akaike’s information criterion (AIC), and conditional maximum likelihood estimator have been used to select the best-fit model for each location. The correlation coefficient between observed and predicted values for all the locations ranges from 0.77 to 0.93. The root mean squared error (RMSE) was found to lie in 0.04 to 0.16 whereas the mean absolute percentage error (MAPE) was observed in 0.06 to 0.2. A comparison of the proposed KSARMA model with the recently developed beta seasonal autoregressive moving average (βSARMA) and Kumaraswamy autoregressive moving average (KARMA) models has also been facilitated. Error estimates such as root mean squared error (RMSE) and mean absolute percentage error (MAPE) reveal that the KSARMA model outperforms the βSARMA and KARMA models except at the Kanyakumari location.

KEYWORDS:

• Wind speed prediction
• Kumaraswamy distribution
• non-gaussian, KSARMA

Acknowledgements

The acknowledgment is duly expressed for the financial grant, presented in the form of a fellowship to the primary author by the Council of Scientific and Industrial Research (CSIR), India.

Disclosure statement

The authors affirm the absence of any conflicts of interest pertaining to this work.

Additional information

Notes on contributors

Mohammad Shad

Mohammad Shad is a Ph.D. research scholar in the Department of Mathematics at the National Institute of Technology, Hamirpur, India. He received his Master’s degree in Mathematics (M.Sc.) from DDU Gorakhpur University, Gorakhpur, India in 2015 and a Bachelor’s degree in Mathematics (B.Sc.) from DDU Gorakhpur University, Gorakhpur, India in 2013. His current research interests are time series modeling and forecasting, machine learning, and deep learning.

Y. D. Sharma

Y. D. Sharma is currently a Professor in the Department of Mathematics at the National Institute of Technology, Hamirpur, India. He received his M.Sc., M.Phil., and PhD degrees in Mathematics from Himachal Pradesh University, Shimla H.P. He has about thirty years of teaching and research experiences in different areas like; thermal, bio, nano convection problems, hydrodynamic/ hydro-magnetic stability, fluid flow, miscible/ immiscible displacement of fluids and wave propagation at solid liquid interface etc.

Pankaj Narula

Dr Pankaj Narula is working as an Assistant Professor in the School of Mathematics (SOM), Thapar Institute of Engineering & Technology (TIET), Patiala. He received his Master’s degree in Mathematics (M.Sc.) from Panjab University, Chandigarh in 2009 and Doctor of Philosophy (PhD) in Applied Mathematics from Indian Institute of Technology Mandi (HP) in 2018. His research interests are mathematical modeling and data analysis.