Wind farm operation trend prediction based on Prony algorithm

The wind farm operation trend prediction model based on machine learning holds significant importance. Due to the randomness and intermittency of wind resources, the output power of wind turbines is difficult to predict, posing challenges to grid stability. This paper proposes the use of a principal component analysis long short-term memory (PCA-LSTM) model to predict the interaction trends between wind farms and the grid and applies the Prony algorithm to analyze the oscillation modes of wind turbine-grid interactions. The research results show that the predicted main oscillation frequencies include 12Hz, 25Hz, and 50Hz, which closely match the actual data with a frequency error of less than 0.01Hz. Additionally, the predicted amplitude results are also highly consistent with the actual values. These findings demonstrate the effectiveness of the method in predicting turbine-grid interaction trends, helping to ensure the safe and stable operation of the grid. By forecasting SSO and SSR in advance, this study provides technical support for the stability prediction and management of high wind power penetration systems.