“…Kamath [16], for example, has analyzed historical wind power data to organize the characteristics of the ramp events in relation to time (duration) and intensity. Utilizing such historical data sets plays an important role in ramp event prediction; therefore various data driven frameworks have been studied [17]- [24]. Several studies have suggested that the application of machine learning methods [25], like auto regressive model [21], support vector machine [17], [18], hidden Markov model [19], random forest [17], [24], gradient boosted trees [23], wavelet transform [21], [26], and artificial neural networks [17], [21], [22], [27], particularly containing deep architectures [28]- [30], contributes to improving the accuracy of ramp event prediction.…”