Power system stability analysis is a key factor in ensuring the safe operation of the power system. With the deepening integration of renewable energy sources and the escalating structural complexity of modern power grids, fast and accurate stability prediction has become particularly important. Traditional model-based methods have the advantage of physical interpretability, but they face limitations such as high computational complexity and insufficient generalization when dealing with high-dimensional and imbalanced data. This study develops a fusion approach that integrates LightGBM and CatBoost. In the data preprocessing stage, categorical variables are transformed through one-hot encoding, and the SMOTE technique is applied to mitigate the class imbalance problem. Subsequently, LightGBM and CatBoost models are trained separately, and their prediction outputs are fed into a logistic regression model as a meta-learner to build the final fusion classifier. Case study findings indicate that the proposed fusion model delivers improved accuracy and F1-score on the training as well as the testing sets, compared with individual GBDT models. Among them, the combination of LightGBM and CatBoost performs best, enabling efficient and accurate identification of transient stability states in power systems.
Enqi Zhang (Tue,) studied this question.