Analysis of wind farm participation in the frequency regulation market considering wind power uncertainty

The participation of wind farms in the bidding of the energy and frequency regulation (FR) markets does not only improve the revenues of wind farms, but is also beneficial to the stable operation of grid dispatching. An optimization model for the bidding plan of a wind farm in the energy and FR markets (E&FR) considering wind power uncertainty is proposed. The wind power uncertainty is described by the probability density function (PDF) prediction model. It integrates the improved extreme learning machine (KELM), a PSO algorithm, and kernel density estimation (KELM-PSO-KDE). The model combines the strong fitting ability of KELM and the non-parametric KDE method, making the PDF prediction results more accurately. Based on the PDF prediction results and the goal to maximize the wind farm revenue, the optimal bidding plan is obtained by particle swarm optimization (PSO). A case study is conducted by using the operating data of a wind farm. The results show that the bidding plan for the wind farm participating in the E&FR markets achieves the maximum revenue and reduces the waste of wind resources, unlike the wind farm that only participates in the energy market. Comparisons with different methods demonstrate the excellent performance of the KELM-PSO-KDE model and the PSO algorithm. This study provides a new method for wind farms to participate in the FR of China’s electricity system.