Subsynchronous Oscillation Identification Method Based on Unthresholded Recurrence Plot and Convolutional Neural Networks

With the integration of a large number of new energy sources such as wind power and photovoltaics plant into the grid, the power system is developing towards a "double-high" trend. The problem of subsynchronous oscillations (SSOs) is becoming increasingly prominent, posing a serious threat to the security and stability of the power system. It is necessary to achieve timely and rapid detection of SSO events. However, the current identification methods based on phasor measurement units (PMUs) suffer from problems such as high computational complexity and difficulties in extracting oscillation characteristics. In view of this, a SSO identification method based on unthresholded recurrence plot (URP) and convolutional neural networks (CNNs) is proposed. This method analyzes the significant feature differences between phasor data during SSO occurrence and normal operation. It extracts and analyzes the recursive features of the data using the URP and converts the features from one-dimensional data sequences to two-dimensional feature maps. A subsynchronous oscillation image recognition method based on CNN is proposed. Furthermore, the method is tested and analyzed using measured data of SSO events to verify its effectiveness and fast response. And it is compared with other identification methods to demonstrate its superiority. The proposed method has shorter recognition window length and has the characteristics of fast recognition and high recognition accuracy.