The invention relates to a power system transient stability prediction method combining manual feature and residual network, and belongs to the technical field of stable analysis of electric power systems. According to the method, all variables of a generator after fault removal are collected, and an initial feature vector is formed; the artificial features are extracted from the initial featuresaccording to human setting, and the initial feature vector is arranged into three-dimensional data, the feature is automatically extracted by utilizing a residual unit in a depth residual network, andthe automatic extraction feature and the artificial feature are jointly used as the input of a full-connection layer, two-layer full-connection layer treatment is carried out to obtain transient stability prediction output to form a structure of a transient stability prediction model; and finally, a training sample set and a verification sample set are adopted, and a relatively optimal model parameter is obtained through iteration solving, so that a final transient stability prediction model is obtained and is used for transient stability prediction. According to the method disclosed by the invention, the prediction accuracy of transient stability of the electric power system can be improved through the combination of the manual characteristic and the residual unit in the depth residual network and the preferred selection of the model parameters.