Stepwise inertial intelligent control for wind power frequency support based on modified stacked denoising autoencoder

Abstract

Due to the large-scale popularization of renewable energy, the rotational inertia of modern power system is reduced, which increases the risk of power system failure and power outage when disturbance events occur. The wind turbine (WT) must be involved in frequency control to respond to power imbalance in the grid when it is detected. Stepwise inertial control (SIC) is one of the important frequency modulation strategies, but it will bring about secondary frequency drop (SFD). In order to minimize SFD while ensuring excellent frequency modulation effect, it is necessary to solve the optimal parameters of SIC, and the traditional method is very time-consuming to solve the parameters in different scenarios. In order to obtain the parameters quickly and accurately when the disturbance events occur, a stepwise inertial control of frequency modulation for wind power based on modified stacked denoising autoencoder is proposed in this paper. First, the sparrow search algorithm is used to obtain the required data. Then, the network model of the modified stacked denoising autoencoder is constructed. The “pre-training, fine-tuning” method is used to train the network parameters, and the Adamax optimization method is introduced to fine-tune the network parameters, which improves the training effect. Finally, the optimal SIC parameters are predicted by adding the scene data into the model. The proposed method is tested in IEEE 9 bus test system. Compared with the traditional method, the proposed method has faster computing speed, excellent prediction accuracy and generalization ability.

Keywords

• Stepwise inertial control
• Secondary frequency drop
• Sparrow search algorithm
• Modified stacked denoising autoencoder
• Adamax optimization