Data-Driven Regulation Reserve Capacity Determination Based on Bayes Theorem

Abstract

To counteract the real-time power fluctuations and maintain the performance of the frequency regulation, it is essential for the system operator to properly determine the frequency regulation reserve capacities (FRRCs). This letter develops a new data-driven method to quantify the FRRCs considering the time-varying wind, solar power outputs, and load power variations. This method mainly includes three steps: first, the concerned power variation ranges are forecasted by using the extreme learning machine-based interval prediction method; second, an adequacy criterion is proposed based on the conditional probability of reaching a certain frequency control standard under a given FRRC and the forecasted power variation ranges; and third, the minimum FRRC satisfying the proposed criterion is determined as the FRRC requirement. To make the high-dimensional probability calculation tractable, Bayes theorem is adopted to simplify the original conditional probability function. The simulation results show that the proposed method can reduce the FRRC and improve the frequency control performance compared with the actual historical data.