Optimized Wind Farm Power Scheduling Using Hybrid Battery Energy Storage Systems

The stochastic nature of high wind power ramp rates is undesirable for stable grid operation. To improve the intermittency of wind power ramp rates, energy storage systems are required. However, a single type of energy storage system and existing battery-supercapacitor systems are inefficient at handling power, energy requirements, and storage health. To solve this problem, this paper proposes a hybrid battery energy storage system with dedicated power-and energy-dense batteries. The novelty of this paper lies, first, in the development of nine different machine learning models and comparing the best method for short-term (15-minute) wind power forecasting in a French wind farm using historical data. Second, the optimizer schedules the optimal power dispatch to the grid based on forecast data. Third, the power distribution control allocates the power between the power-and energy-dense batteries. Fourth, the proposed system is compared with a single type of battery storage system and an existing battery-supercapacitor hybrid energy storage system. Among other models, the support vector regression model achieves an accurate forecast with a coefficient of determination of 96.37%, mean absolute error of 47.18, and root mean square error of 90.13. The results indicate the proposed system not only mitigates the 12% ramp rate limit but also improves the switching transition of batteries from charging to discharging and discharging to charging as compared to a single type of battery energy storage system. When compared to the battery-supercapacitor with the proposed system with distribution power control, the proposed system performs better to improve unnecessary battery utilization.