The dynamic swing angles of suspended insulator strings of transmission lines with different parameters including conductor type, initial stress in conductor, span length and height difference, under stochastic wind filed are numerically simulated by means of the finite element method. Based on the finite element simulation results and the BP neural network, a prediction model for the swing angle is constructed. In the model, the conductor type, span length, height difference, initial stress, wind speed and guarantee factor are taken as the input parameters, and the swing angle as the output parameter. The model is optimized by machine learning and accuracy evaluation with specific evaluation indicators. Swing angle of suspended insulator string in stochastic wind field can be predicted conveniently and fast with this model, which provides a basis for the insulation design of tower head in transmission line.