Prediction of voltage stability margin in power system based on extreme gradient boosting algorithm

The extreme gradient boosting (XGBoost) algorithm was applied in power system voltage stability assessment problem. According to the characteristics of the voltage stability problem, a feature set which could reflect the state of a power system was defined. Taking the absolute value of voltage stability margin as the mapping target, the method to generate the sample set was studied. Based on the introduction of the basic principle of the XGBoost algorithm, the technical details of the algorithm were discussed. The algorithm was evaluated in the IEEE-39 power system. Results show that the XGBoost algorithm has better performance than other machine learning models according to two evaluation metrics: R squared value and mean absolute percentage error value, and has the fastest computation speed, which can meet the demand of online application. Meanwhile, the XGBoost algorithm is proved to be robust when the data errors and data missing happen. And data supplement can be taken for the samples with large prediction deviation to update the model, thus making the performance of the model more stable. © 2020, Zhejiang University Press. All right reserved.