Abstract: Regarding to high dimensionality in power system transient stability assessment,an original feature set was set up,which is irrelevant to system scale. A dual-stage feature selection method based on support vector machine was proposed. During the first stage,the original features are sorted using support vector machine recursive feature elimination method and removed of those behind. Then a group of dimension reduction features is gained. For the second stage,the wrapper method that uses radial basis function kernel support vector machine as classifier is adopted,and a near-optimal feature subset is obtained through best-first search. Finally,in New England 39-bus test system and IEEE 50-generator test system,the feature selection approach was applied in original feature sets and simulation results approved the approach’s effectiveness. Meanwhile,the feature subsets obtained by the dual-stage feature selection,are also valid for other transient stability assessment models.