Abstract: The unbalanced tension and wire jumping caused by the ice-shedding of overhead lines may lead to accidents such as inter-line flashover and tower collapse, seriously affecting the safe operation of the power grid. The prediction method for the response characteristics of isolated-span overhead lines after ice-shedding based on machine learning and finite element simulation. Firstly, the finite element simulation model of isolated-span overhead lines is established and validated. Seven model input parameters are determined. Six key output parameters reflecting the electrical and mechanical characteristics of overhead lines are obtained through feature extraction. Secondly, numerical simulation methods are used to obtain the output parameters of overhead lines after ice-shedding under different input parameter conditions, and the sample dataset is constructed. Then, four types of regression prediction models including BP, RBF, SVM, RF are established and machine learning is used to predict the key output parameters of overhead lines after ice-shedding. Finally, the optimal prediction algorithm for response characteristics parameters of isolated-span overhead lines is obtained. The results show that the RF regression prediction algorithm is the optimal regression prediction algorithm with R² values greater than 0.92 for predicting vertical jump height. The MAE and MSE of unbalanced tension are the best among the four types of regression prediction algorithms, indicating the accuracy of this regression prediction method. The proposed method for predicting the response characteristics of isolated-span overhead lines after ice-shedding can conveniently and quickly achieve the prediction of ice-shedding response characteristic parameters, which will reduce the risk of accidents caused by ice-shedding of transmission lines, and provide reference for the formulation of ice-shedding strategies and emergency plans for transmission lines in heavy ice areas.