At present, the ship-borne wave compensation platform is widely used. However, there is a problem of compensation time delay, which leads to poor compensation effects and potential safety hazards. By predicting the future motion of ships, the operations and controls can be implemented in advance, and the safety of offshore operations can be improved. In this study, the SSA-BiGRU-Attention integrated model was established to predict the future rolling motion of the ship. Firstly, this study used a wave simulation platform to simulate ship motion data. Secondly, the model uses a bidirectional BiGRU neural network to capture critical information from multiple angles and then optimizes network parameters intelligently through the Sparrow search algorithm to improve the efficiency of parameter configuration. Through the improved attention mechanism, the mutual influence between different degrees of freedom is comprehensively analyzed to provide more comprehensive information for prediction. Finally, a multi-step prediction experiment is carried out based on the single-step prediction, and the integrated prediction model is compared with the single prediction model. The mean absolute error and root mean square error of the integrated model can be reduced by 32.68% and 32.48%, respectively, verifying the effectiveness of the integrated prediction model.