As traditional algorithm such as SVM is time-consuming and unable to meet the requirements of assessment online, this paper proposes an extensive kernel core vector machine (ECVM) based on-line stability assessment algorithm of large-scale power system. Firstly, the original feature is extracted based on the decision tree algorithm and then a quick assessment conclusion is given based on ECVM classifier. This algorithm simplifies the new globe calculation process in the minimum enclosing ball issues to avoid the QP problem resolved at each iteration so that the complexity of algorithm is reduced. The simulation results in the New England 39-bus system and a real power system show the superiority of the proposed algorithm, which provides a new idea for the online stability assessment of large-scale power system.