A Method of Frequency Security Margin Index Prediction of Post-disturbance Power System Based on Classification and Regression Model of XGBoost Algorithm

In recent years, power system failures have caused widespread power outages, resulting in huge economic losses. Quickly and accurately predicting the degree of frequency stability of the power system after disturbance is of great significance to ensure the stable operation of the power system. In order to realize the rapid prediction of frequency stability problem, this paper proposes a method for predicting security margin index of frequency based on XGBoost (Extreme Gradient Boosting) algorithm of post-disturbance power system. Firstly, by calculating the integration of the frequency offset for time as a security margin for frequency and using this as an indicator to quantify the degree of frequency stability after a disturbance. Then using the data before and after the disturbance to construct an initial feature set for frequency prediction. And through the Pearson correlation coefficient method for key feature screening. Then, XGBoost algorithm is used to construct classification and regression models, which are combined to predict the frequency security margin index of the post-disturbance system. Simulation analysis of the proposed method is performed in the New England 10-machine 39-bus system. The effectiveness and superiority of the method proposed in this article is verified by comparing it with three machine learning algorithms, BP (Back Propagation) neural network, SVR (Support Vector Regression) and XGBoost algorithm only with regression model as well as one deep learning algorithm, CNN (Convolutional Neural Networks).