Multi-Modal Container Flow Prediction in Sea-Rail Intermodal Transport: A Hybrid Sarima-LSTM Approach

Multi-modal container flow prediction in sea-rail intermodal transport presents significant challenges-from limited historical data and fragmented information systems to difficulties capturing non-linear trends. In addition, due to the absence of a robust operational mechanism and an effective organizational framework, the share of multimodal transport cargo volume in China remains far below that of developed nations. In this paper, we focus on container throughput in searail intermodal transport to enhance forecasting accuracy and optimize resource allocation at ports and rail networks. A hybrid SARIMA-LSTM model is proposed, where the SARIMA component captures long-term trends and seasonal variations, and the LSTM network learns the complex, non-linear dynamics hidden in the residuals. Experimental results demonstrate that the hybrid model significantly outperforms individual models in terms of MAE, RMSE, and MAPE, confirming its effectiveness and stability in handling complex time series data. These findings provide valuable insights for sea-rail intermodal transportation decision-making, helping to improve logistics efficiency and overcome current operational challenges.