In order to achieve intelligent prediction of the overall operation conditions of the hydropower plant structure vibration and solve the problems of the coupling of multiple vibration sources and randomness of vibration response, we proposed a prediction model based on improved genetic algorithm- back propagation neural network (IGA-BPNN) to study the powerhouse vibration response of a riverbed hydropower station. Firstly, the initial weight values and thresholds of BPNN are optimized by IGA, which is characterized by the advantages of efficient parallelism and global search, and then the predicted values of structural vibration displacement are procured by training the BPNN network. The prototype observation example shows that the maximum relative error of the predicted vibration displacements at the measurement points does not exceed 11%; the IGA-BPNN model is significantly superior to other models in terms of prediction accuracy and convergence performance, indicating that the prediction method is effective and feasible. This study can provide a reference for vibration research of other types of hydropower plants.