A modified GM(1, 1) model to accurately predict wind speed. (February 2021)

• Record Type: Journal Article
• Authors: Yousuf, Muhammad Uzair; Al-Bahadly, Ibrahim; Avci, Ebubekir
• Is Part Of: Sustainable energy technologies and assessments. Volume 43(2021)
• Publication Date: 2021-02
• Subjects: Wind speed -- Forecasting -- Grey prediction model -- Remnant -- Developing coefficient; Renewable energy sources -- Periodicals; Energy development -- Technological innovations -- Periodicals; Electric power production -- Periodicals; Energy storage -- Periodicals; 333.79
• DOI: 10.1016/j.seta.2020.100905
• Languages: English
• ISSNs: 2213-1388

Abstract

Graphical abstract: Highlights: A modified grey prediction model is proposed for very short-term wind speed forecasting. The traditional model is modified by L' Hopital's rule and remnant model to fulfill the necessary and sufficient conditions. The proposed model improved the forecasting performance of the traditional model by 9%. The modified model has excellent index of agreement with no unrealistic wind speed predictions. Abstract: Grey prediction models are suitable for uncertain systems and are recognized as a versatile wind energy forecasting technique. However, the traditional model has a disadvantage of seldom failure of necessary conditions. The first-order grey model with one variable [GM(1, 1)] predicts the negative values of wind speeds, which is physically impossible. Also, forecasting results of the traditional model demonstrated that approximately 5% of the predicted values failed to achieve a predetermined accuracy level. In this study, a comprehensive modified GM(1, 1) model is proposed considering wind speed data of Palmerston North, New Zealand. Remnant model and L' Hopital's rule are incorporated to overcome the issues of the traditional method. Results showed that the modified GM(1, 1) model improved the forecasting validity of the traditional model by 98% while the individual accuracy level by 86%. Also, the forecasting performance of the new model is 9% higher than the traditional model. The robustness is further demonstrated by applying the model to three case studies. Overall, the modified model has excellent index of agreement with no negative wind speed predictions.