Wind Power Deviation Charge Reduction Using Time Series Models

Abstract

Globally, power systems are experiencing very high renewable penetration to meet exponentially increasing electrical load demand and reduce carbon emissions. Wind energy has got worldwide acceptance due to its low-cost technologies and widespread availability. However, the high uncertain variability of such wind power generation resulted in various operational, economical, and security challenges such as voltage collapse, wind power curtailment, and frequent system imbalances. These challenges are mainly addressed by deploying additional spinning reserves and specific ancillary market products such as flexible ramp products. This leads to increased system operation costs and such increased costs are usually transferred to Wind Power Producers (WPPs) as penalties. These penalties are called Deviation Charges (DCs) as it is caused by the deviation of actual generation from forecasts. DCs reduce the profit margin of WPPs. DCs are reduced by enhancing forecasting accuracy. In this context, this paper presents a very short-term ARIMA-based time series forecasting model for reducing DCs. ARIMA models are suitable for very short time frame forecasting as they simultaneously use auto-regression and moving average approaches. Proposed ARIMA model forecasts are compared with forecasts obtained from a multi-linear regression model. Results show that the proposed model can produce accurate forecasts and is suitable for reducing DCs compared to the multivariable linear regression model.