ARIMA based statistical approach to predict wind power ramps

Abstract

Wind power penetration in power systems is significantly increasing over the years. Wind generation is highly random and a significant change in wind power within a short timeframe forms a wind ramp event. These events can create severe generation-demand imbalance and cause damage to the wind turbines due to extreme stresses. Therefore, prediction of wind ramp events is essential for system operators to operate the power systems in a secure and reliable fashion. The existing approaches broadly predict based on classification of ramp events, which does not offer efficient results. This paper proposes Autoregressive Integrated Moving Average (ARIMA) based approach for wind ramp predication. Proposed approach is implemented on wind farm located at Bishop and Clerks, Massachusetts, USA to show annual and seasonal distribution results for up and down ramps. Proposed approach is validated through comparative analysis of ramp events obtained using forecasted and actual ramps. The approach is especially effective for short time horizon, offering low error percentage.