Throughput Prediction Of Typical Ports Around The Bohai Sea Based On Long And Short-term Memory Neural Networks

With the gradual improvement of China ’s “Belt and Road” strategic deployment and the continuous deepening of economic globalization,trade between countries is becoming more frequent.As a shipping hub for international trade and a key node in maritime cargo transportation,ports play an important role in international trade.Port throughput has gradually become an important indicator of national competitiveness.The deepening of trade exchanges brings great opportunities for the development of China’s coastal ports,but at the same time,it also puts forward higher requirements for port construction.Therefore,the prediction of port throughput has gradually received great attention from researchers and relevant government departments.This paper analyzes the development status of Tianjin,Dalian,and Qingdao ports and the factors affecting port throughput through typical ports around the Bohai Sea.Based on the integration of the economic data of the three hinterland cities,the LSTM algorithm is used to build a port throughput forecast.The model uses different dimensions of port intelligence data as model inputs to conduct target port throughput prediction experiments to obtain a higher throughput prediction accuracy model.It systematically analyzes and experimentally studies the throughput changes of typical ports around the Bohai Rim.The regional typical port level and the overall regional level of the Bohai Rim port group provide development countermeasures.The specific research is as follows:First,in view of the current situation of port throughput affected by many factors,by analyzing the existing literature on port throughput influencing factors,a relatively reasonable and complete system of port throughput influencing factors is integrated as auxiliary information for throughput prediction.Provide a basis for improving the accuracy of prediction results.Secondly,the wavelet decomposition transform is used to obtain a multi-dimensional auxiliary intelligence data set that includes both low-frequency trends and high-frequency disturbance information.From this,an LSTM-based auxiliary intelligence prediction model is established,and the experimental research of port throughput to predict the development trend of auxiliary information is conducted for the port throughput Prepare for prediction experiments.Thirdly,a port throughput prediction model based on LSTM is separately constructed,and the wavelet decomposition transform is also used to obtain a port throughput intelligence data set that includes both high and low frequency information.The target port throughput prediction experiment is performed through a combination of port dimension data of different dimensions.The model with the highest accuracy is used as the final port throughput prediction experimental model.The empirical comparison of the prediction results under other algorithms(BP,RNN)proves the applicability and accuracy of the LSTM algorithm model in port throughput prediction.At the same time,the analysis of the port development trend based on the forecast results provides a basis for the planning and construction of typical ports around the Bohai Sea.Finally,according to the development status and port throughput trend analysis of typical ports around the Bohai Rim,the development countermeasures for typical port levels in the Bohai Rim region and the overall regional level of the port group in the Bohai Rim region are presented for the challenges facing the development of ports in the Bohai Rim region.