Abstract
The repercussions of high levels of environmental pollution coupled with the low reserves and increased costs of traditional energy sources have led to the widespread adaptation of wind energy worldwide. However, the expanded use of wind energy is accompanied by major challenges for electric grid operators due to the difficulty of controlling and forecasting the production of wind energy. The development of methods for addressing these problems has therefore attracted the interest of numerous researchers. This paper presents an innovative method for assessing wind speed in different and widely spaced locations. The new method uses wind speed data from multiple sites as a single package that preserves the characteristics of the correlations among those sites. Powerful Waikato Environment for Knowledge Analysis (Weka) machine learning software has been employed for supporting data preprocessing, clustering, classification, visualization, and feature selection and for using a standard algorithm to construct decision trees according to a training set. The resultant arrangement of the sites according to likely wind energy productivity facilitates enhanced decisions related to the potential for the effective operation of wind energy farms at the sites. The proposed method is anticipated to provide network operators with an understanding of the possible productivity of each site, thus facilitating their optimal management of network operations. The results are also expected to benefit investors interested in establishing profitable projects at those locations.