In modern power system, traditional transient stability assessment(TSA) methods undergo great challenges as the time domain and space structure complexity continue to increase. Taking into account the massive features generated by the power system, in order to avoid the dimensionality disaster problem in artificial intelligence methods and machine learning models, this paper proposes a novel feature selection method. Based on interaction gain, this method measures both the effectiveness and combination effects of certain feature subset, thereby simplifying the original input without information loss. Case study on IEEE 39-bus system TSA verifies the validation in accuracy, false alarms, calculation efficiency and feature size.
Keywords transient stability assessment, machine learning, information gain, feature subset, AI