Frequency estimation in wind farm integrated systems using artificial neural network

• We model a three-layer neural network using an improved elastic back-propagation algorithm.
• We verify the effectiveness of the method with actual wind farm data.
• The forecasting curve can respond to the change in system frequency in real time.
• We compare the improved method with the other two algorithms in training times and error.

Owing to the random features of wind generator outputs, power system frequency becomes increasingly variable when considerable wind farms are integrated into a power system. Frequency forecast in a power system is difficult and challenging. This study develops an improved elastic back-propagation neural network method to forecast system frequency. The effectiveness of the proposed method is verified using field data from a real wind farm in Guangdong, China. Simulation results show that even in different types of wind farms, the proposed method is applicable and can accurately identify changes in system frequency. Therefore, the forecast results can be used to design appropriate frequency control strategies and to enhance security and stability for the whole system.