Research On Volume Of Road Cargo Transportation Forecasting Method Based On Deep Learning

Highway transportation is one of the main ways of modern transportation,which plays an important role in the whole transportation system.Highway freight volume is the core index to reflect the development of highway transportation and reflect the development of highway transportation.It is of great significance for the further development of road transportation industry to carry out the research on the forecast of road transport cargo volume,the forecast results can not only be used as a reference for highway transportation network planning,it can also provide some data support for the relevant decision of the transport department.At present,the forecast of highway freight volume is mainly realized by time series analysis and forecast method,and there are some shortcomings in accuracy.Therefore,this paper applies the method of Deep Learning to the research of highway freight volume prediction,and tries to obtain better prediction results.In this paper,analysis the characteristics of road freight transport demand combined with the root causes of freight demand at first,point out the necessity of developing and perfecting the theory and method of cargo volume prediction,summarized the related methods for forecasting the demand for freight transportation,includes the traditional prediction method and the new method,that is the deep learning method.The monthly data of Beijing road cargo volume from 2005 to 2017 are selected as raw data.The traditional time series method,the LSTM model based on the Deep Learning and the ARMA-LSTM combination model are used to realize the prediction of the traffic volume of the highway goods and the prediction results are obtained.The prediction of LSTM model based on Deep Learning is divided into two parts.At first,considering that the cargo volume may be affected by seasonal factors,the cargo volume data are classified according to the monthly,and the data information of each month cargo volume is used to realize the forecast of the cargo volume data in each month.The second is to ignore the effects of seasonal factors,using all historical data of cargo traffic to realize the forecast of monthly cargo traffic in 2017.The combined model is used to predict the volume of cargo by applying the two models to a complete forecasting process at the same time.In this paper,the prediction process of LSTM model in deep learning is realized with thehelp of Matlab platform.The validity of the model was evaluated by selecting RMSE(root mean square error)and MAPE(average absolute error percentage).Taking the forecast of highway freight volume in Beijing as an example,the prediction of the model is verified.The results show that the ARMA-LSTM combined model can fully reflect the variation law of the freight volume data in the forecast of highway freight volume,and has good reliability and effectiveness.We can try to apply this method to the forecast of highway freight volume in other provinces,and provide the support for analyzing the development trend of highway freight volume effectively.