The invention belongs to the field of smart power grid safety, and discloses a smart power grid node vulnerability assessment method, system and device, and a storage medium, and the method comprises the steps: obtaining the network topology of a smart power grid; according to the network topology of the smart power grid, obtaining the association importance, the position importance, the mediation centrality, the network connectivity loss under a node deletion method, the node embedding vector and the expansion mediation centrality of each node of the smart power grid; according to the association importance degree, the position importance degree and the mediation centrality of each node of the smart power grid, the network connectivity loss under a node deletion method, the node embedding vector and the expansion mediation centrality, constructing a feature vector of each node of the smart power grid; and inputting the feature vector of each node of the smart power grid into a preset vulnerability analysis model based on a machine learning classifier to obtain the vulnerability of each node of the smart power grid. The analysis accuracy of the node vulnerability is greatly improved, the method does not depend on a complex fault model any more, the expansibility is very high, and the calculation complexity is reduced.