Short-Term Voltage Stability Assessment of Multi-infeed HVDC Systems Based on JMIM and XGBoost

摘要

The intensive infeed of multiple large-capacity HVDCs into the receiving-end power system has introduced a significant challenge on short-term voltage stability (STVS) management. This brings STVS evaluation a necessity to enable enhanced system evolution performance against the risk of blackouts. To tackle this issue, this paper proposes a fast and accurate STVS assessment approach based on Joint mutual information maximization (JMIM) and eXtreme Gradient Boosting (XGBoost). JMIM efficiently selects crucial input features from the raw features with high dimensions, thereby reducing the complexity of the model and avoiding the dimension explosion issue. Aided by the second-order Tailor expansion and the regularization term, improved STVS assessment performance can be achieved via XGBoost. Simulation results on the modified New England 39-bus system demonstrate the superiority of the proposed approach over some state-of-the-art machine learning algorithms.