Global Horizontal Irradiance Forecast at Kanto Region in Japan by Qunatile Regression of Support Vector Machine

In the interests of the stable operation of the transmission system, transmission system operators (TSOs) procure regulating power supplies to cope with significant deviations from renewable energy forecasts. Therefore, it becomes important to improve the average precision of the one-day ahead forecast and to decrease the maximum error of the forecast in a power transmission system with a large number of photovoltaic systems. In this paper, the quantile regression using support vector machines is applied to the prediction of the previous day's solar radiation, and it is confirmed that maximum width of the error can be reduced while suppressing the minimum length of the prediction error.