Toward Intelligent Inertial Frequency Participation of Wind Farms for the Grid Frequency Control

Evolving dynamics of modern power systems caused by high penetration of renewable energy sources increased the risk of failures and outages due to declining power system inertia. Large-scale wind farms must participate in frequency control that responds optimally in due time and adaptively in case of detecting power imbalance in the grid. Existing research studies have shown interest on stepwise inertial control (SIC) on wind turbines (WTs). However, the adequate power increment and time duration of WTs using SIC are the key questions that have not yet been fully addressed. This paper proposes an intelligent learning-based control system for WTs participation in frequency control, as well as for mitigating negative effects of the SIC. First, an appropriate optimization model for grid frequency control is defined. Then, the model is solved using lightning flash algorithm (LFA), imperialist competitive algorithm, and particle swarm optimization to control the WTs in a wind farm. The obtained dataset by LFA are applied to an artificial neural network that is trained with the Levenberg-Marquardt algorithm and LFA. The proposed control system optimally adjusts the power increment and duration time of participation for each WT in the farm. Analyses on a 100 MW wind farm has been integrated into the IEEE 9-bus system and experimental tests have proved the efficacy of the proposed approach.