基于VMD-GWO-LSTM组合模型的港口集装箱吞吐量预测

Forecasting of Port container throughput is of great significance to port construction and regional economic development. In order to improve the prediction accuracy, this paper proposes a VMD-GWO-LSTM combination model, first using the VMD decomposition method to decompose the original throughput sequence into several sub-sequences, and then using GWO-LSTM to forecast each sub-series separately, GWO is used to optimize the number of hidden layer units, learning rate and regularization coefficient of LSTM, and the Forecasting results of the subseries are integrated and added to obtain the final Forecasting results. In order to verify the versatility and Forecasting ability of the model, four ports of Shanghai Port, Ningbo-Zhoushan Port, Qingdao Port and Guangzhou Port were selected as experimental objects, and seven other intelligent Forecasting models were used for comparative experiments. The results show that the VMD-GWO-LSTM combination model performs best in RMSE, R2 and MAE, and is 26% lower than the suboptimal VMD-PSO-LSTM combination model on RMSE and 6.3% better on R2, which is far better than the single LSTM model, with high Forecasting accuracy and fit.