Research On Cargo Throughput Forecast Of Qingdao Port Based On Combination Model

In recent years,ports have gradually occupied an indispensable position in the world trade and logistics system,and port logistics transportation has developed into an inevitable trend in logistics research.Reasonable prediction of port cargo throughput is a key part of port logistics planning and construction of logistics park,which is conducive to giving full play to port effect and promoting sustainable and rapid development.This paper conducts a preliminary theoretical analysis and practical research on the single-item model and combined model of Qingdao Port cargo throughput forecasting.Through continuous improvement of forecasting methods to improve the accuracy and scientificity of throughput forecasting,the purpose is to find an effective model for port throughput forecasting to guide the planning and layout of Qingdao Port Port Logistics.Firstly,analyze the development status of Qingdao Port in the early stage of throughput forecasting.The natural location and port facilities of the four major port areas of Qingdao Port were introduced.In order to grasp the basic trend of Qingdao Port throughput for many years,the rules and characteristics of Qingdao Port’s cargo throughput,container throughput and foreign trade throughput data are fully analyzed.According to the development status of Qingdao port,a detailed SWOT analysis is made to analyze its advantages,disadvantages,opportunities and threats.From six aspects of port hinterland,this part analyzes the factors influencing the cargo throughput of Qingdao port in detail,which lays a foundation for the prediction of port throughput.Secondly,the single-model verification of Qingdao Port’s cargo throughput forecast is carried out.Five key indicators affecting cargo throughput were extracted by principal component analysis before forecasting,and the data in the next five years were forecasted by combination model as input data for subsequent throughput forecasting.Then we use multiple regression model,BP neural network model and ARIMA model to forecast the cargo throughput of Qingdao Port from 2018 to 2022,and make a comprehensive analysis and comparison of the three models.Through the comparison of relative error and fitting curve,we find that the ARIMA model in the single prediction model has the highest prediction accuracy and the best fitting effect,but the prediction effect is not ideal.Finally,the combined model of Qingdao port cargo throughput forecasting is validated.According to whether the weight coefficient changes with time,two kinds of weighted combination models and three variable weight combination models are selected to apply to the forecast of Qingdao Port cargo throughput.After all the model evaluations and multi-faceted analysis,the relative error of the variable weight combination model based on IOWA operator is smaller,the fitting curve is smoother,and the prediction result is more reliable.The rationality of the forecasting results is analyzed from three aspects of overall strategy,economic development and throughput development trend.Combined with the forecast result of Qingdao port’s throughput and the related status of SWOT analysis,this paper puts forward specific suggestions for the logistics development of Qingdao port,and provides an important reference for Qingdao Port transportation development planning.The research shows that the application of the combined model for port throughput prediction is more accurate than the single-item model.The variable-weight combination model provides a new idea for port throughput forecasting.This paper provides a scientific theoretical basis and decision support for the throughput forecasting and port logistics planning of Qingdao Port.